Orphan neuropeptides and receptors: Novel therapeutic targets

1. Introduction

Peptides function as chemical messengers throughout the body as neuropeptides (e.g., enkephalin), as peptide hormones (e.g., insulin), or both hormone and neuropeptide (e.g., oxytocin). Peptides affect a wide range of physiological and behavioral activities, including feeding and body weight regulation, sleep wake cycles and alertness, sexual activity and reproduction, pain, anxiety, fear, stress, depression, and the regulation of homeostasis of many parameters (blood glucose, temperature, etc) (Fricker, 2012; Strand, 2003). Due to the large number of processes controlled by peptides, there has been considerable interest in the development of drugs that target peptide receptors. The most widely used drugs targeting neuropeptide receptors are the opiates, including natural products such as morphine as well as synthetics like fentanyl and oxycodone, all of which are commonly prescribed for the treatment of pain. Antagonists of the mu opioid receptor are used both to treat opioid overdose (naloxone) and to reduce the cravings of opioids, alcohol, nicotine, and food (naltrexone) (Aboujaoude and Salame, 2016; Greig and Keating, 2015; Sudakin, 2016). Other FDA-approved medications that target neuropeptide receptors include aprepitant, which is an antagonist of the substance P/neurokinin receptor NK1 and is used to treat nausea and vomiting, and suvorexant, an antagonist of the orexin receptors OX₁ and OX₂ that is used to treat insomnia (Preskorn, 2014). FDA-approved peptide analogs that target peptide hormone receptors include analogs of oxytocin (carbetocin), vasopressin (desmopressin), and glucagon-like peptide-1 (exenatide and others) (Fosgerau and Hoffmann, 2015; Janzen et al., 2016; Manning et al., 2008). Many more compounds are in development for a variety of indications, including drugs that target corticotropin-releasing factor CRF-1 receptors, neuropeptide Y-1, -2, and -5 receptors, melanocortin MC4R receptors, tachykinin receptors, neurotensin receptors, and others (Barak et al., 2016; Griebel and Beeske, 2012; Griebel and Holsboer, 2012; Jagoda et al., 2011; Kaspar and Reichert, 2013; Zorrilla and Koob, 2010). Most of these targets are G protein-coupled receptors (GPCRs), the largest class of receptors in the mammalian genome.

2. GPCR families and orphan receptors

All known GPCRs can be grouped into five main families based on consideration of amino acid sequence similarities (Fredriksson et al., 2003). The families can be divided into groups and the groups further divided into sub-branches. In most cases these sequence-based divisions correspond to the types of ligands that the receptors recognize. Many of the putative GPCRs identified from analyses of the human genome or other nucleotide sequencing approaches did not have a known function—these were termed “orphan” GPCRs. In order to understand the biological function of orphan receptors, there has been a large effort to identify endogenous receptor ligands and some of these searches were successful (Civelli et al., 2013; Civelli et al., 2006).

Despite considerable efforts, approximately 60 GPCRs in the Family A group (which includes neuropeptide receptors) have not been de-orphanized; their endogenous ligands remain unknown (Khan and He, 2017; Ngo et al., 2016). There are three potential reasons for this. First, it is possible that endogenous ligands have not been detected in the screening assays used because either the screening assay or the conditions used to detect receptor activation were not correct or sensitive enough. For example, the assays often used to deorphanize receptors involve high throughput methods that are easy to use and give a big signal to noise ratio, but which may not be appropriate for a given receptor-ligand combination. High throughput assays are generally carried out using a single dose of ligand at a single time point. Such assays could easily miss the detection of ligands for receptors that are rapidly desensitized at high doses, or which show a delayed response to ligand activation. Also, the assay might not be appropriate in that the ligand might bind to the receptor but activate a distinct signal transduction pathway (Figure 1); this concept is commonly referred to as biased agonism or functional selectivity. Alternatively, the endogenous ligand might bind to only the dimeric receptor activating a distinct signal transduction pathway as compared to individual receptors (Figure 1). Unless multiple assays are used for deorphanization, the identification of the appropriate ligand for a given receptor can be missed.

Second, an endogenous ligand may not exist. GPCRs often form heterodimers, and some members of the GPCR gene family appear to have lost their ability to directly interact with ligands and instead function only in the modulation of other GPCR proteins (Tao and Conn, 2014). For example, γ-amino butyric acid (GABA) binds to and activates the GPCR named GABA-B R1, but does not bind to the related protein named GABA-B R2. The major function of the R2 protein is to interact with the R1 protein—this interaction is important for cell surface expression of the R1-R2 complex (Pin et al., 2004). Alternatively, some GPCRs are activated by proteases rather than by endogenous ligands (Krumm and Grisshammer, 2015).

The third potential reason that an orphan GPCR has not been de-orphanized is that the appropriate ligand was not tested. Some studies tested brain extracts for the presence of a factor that could activate a receptor, and then fractionated the extract until the compound was pure enough to identify (Civelli et al., 2013; Devane et al., 1992; He et al., 2004; Kojima et al., 1999; Reinscheid et al., 1995; Sakurai et al., 1998; Shimomura et al., 2002). In theory, this approach should be able to deorphanize any receptor for which the endogenous ligand is present at levels sufficient to produce a response above background. However, many endogenous ligands are labile and rapidly degraded in brain extracts. Other high throughput studies avoided the problems with brain or tissue extracts and started with purified molecules (Liu et al., 2003; Meyer et al., 2013; Rezgaoui et al., 2006). However, these studies were limited by the input material and would not have detected novel peptides that were not in their libraries. Furthermore, a general assumption was that receptors would bind molecules similar to the families to which the orphan GPCRs belong. For example, the orphan receptor named GPR171 exhibits highest homology to the P2Y family of receptors and was tested with a variety of P2Y ligands such as nucleotides, dicarboxylic acids and lipids, none of which has been found to exhibit detectable binding to GPR171 (Gomes et al., 2013; Wardman et al., 2016). Because of this prediction, GPR171 was probably not extensively tested for binding to neuropeptides, especially novel “orphan” peptides. Just as there are orphan GPCRs, there are orphan neuropeptides—peptides that appear to be neuropeptides based on their cellular and intracellular location, but which do not have known receptor binding partners.

3. Orphan Neuropeptides

Unlike GPCRs, which are evolutionarily related and have regions with conserved amino acid sequences, there are no sequences conserved among all neuropeptides. Thus, novel neuropeptides cannot be identified from amino acid homology and alternate approaches have to be used. In this context, a biochemical approach was used to identify novel neuropeptide precursors in mice lacking carboxypeptidase E (CPE) activity (Che et al., 2001). This enzyme plays an important role in the biosynthesis of nearly all neuropeptides, removing lysine and/or arginine residues from the C-terminus of the peptide-processing intermediates (Fricker, 2012; Fricker and Snyder, 1982; Fricker and Snyder, 1983). Mice with a mutation in the Cpe gene accumulate peptide-processing intermediates (Naggert et al., 1995; Varlamov et al., 1996; Zhang et al., 2008). Together with colleagues, we isolated these intermediates using affinity chromatography selective for peptides with C-terminal lysine or arginine, and sequenced the peptides using mass spectrometry (Che et al., 2001). This led to the identification of a number of novel mouse brain peptides, including peptides we named SAAS, PEN, and LEN (Fricker et al., 2000). These three peptides and several others are derived from the same precursor, proSAAS (Figure 2).

Subsequent studies found proSAAS-derived peptides to be the most commonly detected peptides in mass spectrometry-based peptidomics studies on mouse brain, suggesting that these peptides are among the most abundant brain peptides (Fricker, 2012). All of the proSAAS-derived peptides are highly conserved between mouse and human, with amino acid sequence identities of over 50%; it is likely that high conservation reflects biological activity for these peptides (Table 1). Other orphan peptides that we found many times in peptidomic studies on mouse brain and which are conserved >50% between mouse and human are listed in Table 1. Many other peptides are found in typical peptidomics studies which are not highly conserved. For example, most peptides derived from prothyrotropin releasing hormone (proTRH) detected in mouse brain only show 35-46% sequence identity between mouse and human, and only one proTRH peptide exceeds 50% conservation (Table 1). Peptides with known functions (i.e. thyrotropin releasing hormone, enkephalin) were excluded from Table 1, although fragments of peptides known to be bioactive are included on the list. One example is dynorphin A10-17, which is the C-terminal portion of dynorphin A17 that is released after cleavage into dynorphin A8. Similarly, several highly conserved fragments of proenkephalin that we detected in peptidomic studies are also present as larger forms with N-terminal enkephalin sequences attached (not shown). In these cases, the high conservation may reflect the function of the larger peptides, which bind to opioid receptors. Alternatively, the cleaved peptides may have distinct functions from their precursors. Because the N-termini of the opioid peptides are essential for binding to the opioid receptors, if the cleaved C-terminal fragments have biological activities it is presumably through interaction with other proteins.

The list of peptides shown in Table 1 is not a complete list of likely orphan peptides. For example, the list does not include the well-studied orphan peptide named CART (cocaine and amphetamine-regulated transcript) which is a 41-48 residue peptide that has been reported to function through an unidentified GPCR (Lakatos et al., 2005; Vicentic et al., 2006; Vicentic et al., 2005; Yermolaieva et al., 2001). Although CART was not detected in our peptidomics studies, possibly due to its large size and/or disulfide bonds, several N-terminal fragments of the CART precursor protein (CARTP) were detected (Table 1). Some of the peptides in Table 1 have been reported to have biological activities, such as the peptide derived from cerebellin 1 precursor protein. This cerebellin peptide, and a shorter form missing the N-terminal residue which was also commonly found in peptidomics, have been reported to stimulate the secretion of catecholamines from cultured adrenal cells and to modulate nociception in an in vivo assay (Albertin et al., 2000; Su et al., 2014). While some of these orphan neuropeptides may function as endogenous ligands by directly activating GPCRs, others may function as indirect neuropeptides (described below in section 5).

4. Matching orphan peptides and orphan receptors

5. Complexity of intercellular signaling

The modulation of neuropeptide signaling can occur via multiple mechanisms. One such mechanism is that of biased agonism. Some endogenous neuropeptides have been recently shown to function as biased agonists at GPCRs, differentially activating G protein and non-G protein coupled pathways (Wisler et al., 2014). For example, while Agouti-related protein (AgRP) functions as an MC4R antagonist when examining G protein signaling, it functions as a potent MC4R agonist activating the extracellular signal-regulated kinase pathway (Tao, 2014) and opening the Kir7.1 potassium channel (Ghamari-Langroudi et al., 2015). As discussed above, another mechanism of modulation of receptor signaling is by receptor-receptor interactions i.e. dimerization (Figure 1).

Most peptide precursors are cleaved into a variety of peptides that are co-stored and co-released, and in some cases each of these peptides can activate distinct receptors and signaling pathways (Figure 3). For example, proopiomelanocortin (POMC) is cleaved into αMSH, β-endorphin, and a number of other peptides such as joining peptide. ProSAAS also gives rise to a number of distinct peptides (Figure 2). ProNPY is cleaved to produce NPY and a highly conserved C-terminal fragment (Table 1). ProCART and AgRP are also cleaved to produce multiple peptides (Fricker, 2010; Secher et al., 2016). Since all of these peptides are presumably co-secreted by neurons that express the peptide’s precursor protein, it is likely that they contribute to complexity in the signaling (Figure 3).

When multiple peptides are co-secreted, there are two ways the peptides can interact. For those peptides which bind to receptors, the different signaling cascades can produce positive or negative interactions. Another completely different mechanism is through competition for the extracellular peptidases that cleave the peptides into smaller fragments. In some cases, the smaller fragments have biological activities that differ from the parent peptide, and in other cases the fragments are completely inactive (Figure 4). None of the extracellular peptidases that have been identified are specific for a particular peptide, and all are able to bind a range of different peptides (Cudic and Fields, 2009; Turner, 2003). Thus, the multiple peptides secreted by a neuron will compete for the limited extracellular peptidases, thereby limiting degradation and leading to local increases in their levels (Figure 4). One bioactive peptide that is known to target extracellular peptidases, rather than receptors, is opiorphin (Thanawala et al., 2008; Wisner et al., 2006), and many more examples are likely. This class of peptide has been termed “indirect neuropeptides” (Fricker, 2012). It is interesting to note that AgRP folds into a structure with 5 pairs of cysteine disulfide bridges; this structure is referred to as a “knottin” and has homology to protease inhibitors. CART contains three disulfide bridges and also folds into a compact structure with similarity to cystine knots (Ludvigsen et al., 2001). However, the idea that AgRP and/or CART function in part as protease inhibitors does not appear to have been tested. It is likely that more examples will be found of indirect neuropeptides, in which secreted peptides interfere with the degradation of neuropeptides that directly bind to receptors. It is possible that some of the highly conserved peptides listed in Table 1 as potential orphan neuropeptides are functional in this alternative capacity rather than through directly binding to GPCRs. Conservation of amino acid sequence suggests a function, but this function is not necessarily through direct binding to receptors.

6. Summary and Future Directions

The recent advances in detection of peptides and proteins in various brain regions have led to the identification of a number of peptides and GPCRs. Using a combination of bioinformatics approaches to identify likely targets, followed by medium-throughput screening, we identified GPR83 and GPR171 as receptors for peptides PEN and LEN. Characterization of these receptor systems as well as other receptor systems in the hypothalamus has begun to reveal some of the intricacies of neuropeptide receptor signaling. However, a complication is the redundancy inherent to many biological systems. This is clearly the case for the AgRP neurons that secrete AgRP, NPY, PEN, LEN, and GABA as well as other peptides derived from proAgRP, proNPY, and proSAAS. Deletion of any individual signaling molecule (i.e. knock-out mice) does not show severe phenotypes of body weight alteration, even though these neurons are clearly of fundamental importance in the control of feeding (Gropp et al., 2005; Luquet et al., 2005; Qian et al., 2002). But rather than simply being a case of redundancy of signal pathways, each of the signaling molecules can contribute meaning to the message. Optogenetics and chemogenetic techniques in combination with temporal, tissue or cell-type specific expression are likely to help characterize the contribution of neuropeptides from individual neurons since they allow for specific stimulation of selective neurons. Because the total effect of the multitude of secreted molecules is likely to be greater than the sum of the parts, these approaches can complement studies exploring the function of individual signaling molecules. An important area for future studies is to examine the release of neuropeptides from neurons in vivo, although techniques to accomplish this are not yet able to detect the vast majority of peptides. Another important area is to understand the ways by which the receptor activity at the postsynaptic neuron is regulated and the hierarchy of this regulation. The effect of extracellular proteolytic activity on peptides, and the competition of peptides for these peptidases is another area for further studies. By developing strategies to explore and delineate the complexity of neuropeptide receptor signaling, we will be better equipped to develop pharmacological agents specifically targeting those receptor systems, circuits, behaviors and pathologies.