Beyond CGRP: the calcitonin peptide family as targets for migraine and pain

Abstract

The calcitonin gene-related peptide (CGRP) system has emerged as a key pharmacological target for the treatment of migraine. However, some individuals who suffer from migraine have low or no response to anti-CGRP or other treatments, suggesting the need for additional clinical targets. CGRP belongs to the calcitonin family of peptides, including calcitonin, amylin, adrenomedullin and adrenomedullin 2. These peptides display a range of pro- and anti-nociceptive actions, including in primary headache conditions such as migraine. Calcitonin family peptides also show expression at sites relevant to migraine and pain. This suggests that calcitonin family peptides and their receptors, beyond CGRP, may be therapeutically useful in the treatment of migraine and other pain disorders. This review considers the localisation of the calcitonin family in peripheral pain pathways and discusses how they may contribute to migraine and pain.

Introduction

Chronic pain disorders are among the most prevalent and disabling non-communicable diseases, ranging from tumour-associated pain to lower back and hip pain (Goldberg & McGee, 2011). They result in a reduced quality of life for individuals and are associated with a significant social and economic burden due to direct health care costs and lost productivity (NIH, 2011). Headache, including cluster headache, medication overuse headache, post-traumatic headache, and migraine, can be particularly painful and disabling chronic disorders. The most prevalent of these is migraine, which affects approximately one billion people and is the sixth highest cause of disability, defined using disability-adjusted life years as suggested by the World Health Organisation (Whiteford et al., 2015).

The mechanisms of migraine are unclear but appear to involve sensory nerves analogous to those associated with chronic and acute pain. These sensory nerves are involved in neurogenic inflammation and sensitisation to pain, which may maintain or contribute to chronic pain states. The trigeminal nerve, which is the primary sensory nerve that controls craniofacial pain, is made up of pseudounipolar sensory neurons. Cell bodies of these neurons are located in the trigeminal ganglia (TG) (J. C. A. Edvinsson et al., 2020). Trigeminal nerve fibres innervate the face and transmit sensory information to second order neurons in the dorsal horn of the spinal cord (C1 and C2) and the spinal trigeminal nucleus (STN) in the brainstem (J. C. A. Edvinsson et al., 2020). The sensory neurons that are relevant for body pain have cell bodies located in the dorsal root ganglia (DRG) and transmit sensory information to second order neurons in the dorsal horn of the spinal cord. Second order neurons then transit information up the trigeminothalamic or spinothalamic tracts where they synapse with several brain regions, including the thalamus. The TG and DRG play a major role in the transmission and modulation of nociceptive signals outside of the brain. The ganglia are highly vascularised structures outside the blood brain barrier, allowing them to act as an interface between circulating or peripherally produced substances, including peptides. Many peptides are peripheral neuromodulators, such as substance P, and are important in pain transmission (Zieglgansberger, 2019). This makes peripheral peptides that are produced within, or have access to, the TG and DRG potentially important clinical targets in the treatment of painful disorders.

One such peptide which modulates pain is calcitonin (CT) gene-related peptide (CGRP). CGRP is a member of the CT peptide family, which includes CT and amylin (Hay et al., 2018). CGRP is a sensory neuropeptide with a well-described causal relationship to migraine and a proposed general role in pain modulation (Iyengar et al., 2017). This made CGRP an attractive drug target and led to the approval of small molecule and monoclonal antibody drugs that reduce CGRP activity for the prevention or treatment of migraine and cluster headache (Edvinsson, 2018). These drugs act by binding to either CGRP or a CGRP receptor and prevent CGRP receptor activation. It is not known whether these drugs are effective for other chronic pain disorders. Interestingly, patient responses vary, and these drugs do not offer relief to all migraine patients, suggesting that other factors are involved in migraine pathophysiology. Potential contributors could be the additional members of the CT peptide family that are implicated in nociception and are elevated in conditions involving pain.

There is substantial evidence outlining the role of CGRP in migraine, including a large body of research describing the expression of CGRP in migraine-relevant structures. However recently other members of the CT peptide family have been shown to be elevated in migraine or able to induce migraine attacks. Therefore, the purpose of this review is to describe how each member of the CT peptide family might contribute to pain. We will introduce the CT peptide family members and their receptors. We will then discuss what is presently known about the contribution of each individual peptide to pain and describe their localisation in structures related to pain, including spinal cord and the peripheral sensory nerves of the TG and DRG. Limitations and challenges associated with interpreting the localisation data are also outlined.

Overview of the calcitonin peptide family

The CT family of peptides is involved in a wide range of homeostatic processes and is relevant for a number of diseases, including diabetes, osteoporosis, cancer, obesity, and pain disorders. This peptide family consists of six major members, CT, amylin, adrenomedullin (AM), adrenomedullin 2 (AM2, also known as intermedin), α-CGRP and β-CGRP (Figure 1). These six peptides are the only known members expressed in humans, rats, and mice. However, in other mammals the complement varies and additional peptides have been reported (Katafuchi et al., 2009). There is variability in amino acid length and limited sequence identity for these peptides, but they display overlapping receptor pharmacology. This may in part stem from shared structural features; all family members contain an amidated C-terminus, an N-terminal disulphide bond and α-helical and β-turn elements (Hay et al., 2018).

Figure 1: Amino acid sequences of the human CT peptide family.

In all peptides a disulphide bond is formed between the two N-terminal cysteines and each have a C-terminal amide (not shown). A) The human CT peptide family, omitting the N-terminal extensions of AM and AM2. B) Alignment of full-length AM and AM2. C) Sequence alignment of human α-CGRP and Amy. D) Overlap in amino acids between human and rodent amylin and αCGRP. Alignment performed in Clustal Omega (Sievers et al., 2011) and analysed using BoxShade (http://www.ch.embnet.org/software/BOX_form.html). For A-C, black indicates an exact match, grey indicates 70–100% similarity and white indicates <70% similarity. For D, dark blue indicates that an amino acid is shared across all four peptides; light blue indicates that an amino acid is shared across three peptides. Substantial sequence overlap is observed at the N and C terminus of the peptides (underlined).

Overview of the calcitonin receptor family pharmacology and expression

All members of the CT peptide family activate members of the CT receptor family, a sub-family of Class B G protein-coupled receptors. Although the focus of review is peptides, these receptors warrant introduction because of their heterodimeric nature and complex pharmacology. The CT receptor (CTR) and the CT-like receptor (CLR) interact with receptor activity-modifying proteins (RAMPs) to generate seven pharmacologically distinct receptors. Splice variants of CTR expand this repertoire. CLR forms heterodimers with RAMP1, RAMP2 and RAMP3 generating CGRP, AM₁ and AM₂ receptors, respectively (Hay et al., 2018; Hendrikse et al., 2018). The CTR forms AMY₁, AMY₂ and AMY₃ receptors with each RAMP, respectively. CTR can be expressed at the cell surface alone. The pharmacology of these receptors is complex due to shared receptor components. Therefore, it is unsurprising that there is considerable overlap in agonist and antagonist activity (Figure 2).

Figure 2: The human CT receptor family subunits, receptors, and pharmacology.

The relative activities of agonists and antagonists at human receptors are shown with arrows to indicate at which receptors they are most active. The relatively high potency of CT reported at the human AMY receptors likely relates to the activation of free CTR, not AMY in these systems. Solid arrows indicate relatively potent activity compared to dashed arrows, which indicate weaker activity (Hay et al., 2018). Created with BioRender.com.

The complex pharmacological relationship that exists between the various CT peptides and their receptors highlights the difficulties in attributing the activity of an exogenously applied peptide to a specific receptor-mediated response. For instance, the application of a relatively high or supraphysiological concentration of AM may activate the CGRP receptor and therefore mimic the endogenous activities of CGRP. This may also be of pathophysiological significance as AM is highly expressed in a number of diseases where pain can be a feature, including several cancers (Hay et al., 2011). Receptor antagonists can sometimes be useful because they can be used to block the activity of endogenously expressed peptides to investigate their physiological role. Additionally, they can be employed to elucidate the contribution of individual receptors to the pharmacological effects of exogenously administered peptide. However, they need to be selective and the overlapping receptor pharmacology observed with the CT family agonists is also observed with commonly used receptor antagonists (Figure 2). Care should therefore be taken when interpreting both agonist and antagonist data. Comparing pharmacological data to the anatomical localisation and expression patterns of the peptides and receptor components at their sites of production and action can give further insight into the likely endogenous peptide and receptor involved in a physiological response.

The expression of receptor components in the nervous system has been thoroughly reviewed elsewhere (Hendrikse et al., 2018). In broad terms, the receptor components for several members of the CT receptor family have been reported at sites involved in pain. CGRP receptor components are highly expressed throughout the peripheral nervous systems (PNS) and central nervous systems (CNS), including at the majority of sites involved in pain transmission and processing (Eftekhari et al., 2010; Hendrikse et al., 2018). Interestingly, the AMY₁ receptor, which can also be activated by CGRP (Figure 2), has been observed in regions that are important for pain, such as the TG (Walker et al., 2015). CTR expression is reported throughout the CNS and PNS, with mRNA and protein detected in areas relevant to pain, including the TG, DRG and spinal cord, as well as higher processing centres, such as the thalamus and the amygdala (Hendrikse et al., 2018; Ray et al., 2018; Walker et al., 2015). RAMP1 is also reported to be expressed in several of these locations (Eftekhari & Edvinsson, 2011; Hendrikse et al., 2018; Walker et al., 2015). Additionally, CTR, RAMP2 and RAMP3 expression was described in some of these structures. Therefore, it is possible that AMY₂ and AMY₃ receptors are present (Cottrell et al., 2005; X. Huang et al., 2010). mRNA and immunoreactivity studies suggest that components of the AM₁ and AM₂ receptors are present in the TG, DRG and spinal cord (L. Edvinsson et al., 2020; Hong et al., 2010; Ma et al., 2006; Ray et al., 2018; D. Wang et al., 2013). However, a cautionary approach should be taken to interpreting data with all CLR, CTR and RAMP antibodies, unless thorough antibody validation has been conducted. For most of them, this is not the case.

Challenges for defining the expression of the calcitonin family of peptides in tissue

Understanding the spatial relationship between the CT family of peptides and their receptors is important. It is vital to know whether these peptides are likely to act at their receptors in an autocrine or paracrine fashion from locally expressed peptide, or in an endocrine fashion due to peptide in the circulation. This is key to understanding the pathophysiological role of these peptides in pain and the successful development of pharmaceuticals targeting these peptides and their receptors.

However, investigating these spatial relationships is difficult. Only approximately 40% of mRNA is translated into protein and in neurons, proteins can be transported from distant cell bodies to remote fibres. This may result in an apparent lack of mRNA at a remote site, which leads to the conclusion that the gene is not expressed at this location, despite protein being present (Liu et al., 2016; Tian et al., 2004). Additionally, antibodies frequently display non-specific or off-target binding, making it difficult to interpret the data presented. Ideally all antibodies would be validated using genetic models, including knockout or knock-down models, however this is not always possible for many reasons including model availability (Uhlen et al., 2016).

As previously stated CLR, CTR and RAMP antibodies are poorly characterised, preventing thorough investigation into the distribution of these receptors. Similar problems with tool validity have been reported for peptides. Both mRNA probes and antibodies have displayed cross-reactivity between the CT family peptides (Ahren & Sundler, 1992; Ferrier et al., 1989; Nicholl et al., 1992; Rees et al., 2021; Tingstedt et al., 1999). This is especially relevant when one peptide is particularly abundant at a given location. For example, CGRP is highly expressed in neuronal vesicles, whereas amylin is highly expressed in the β cells of pancreatic islets (Hay et al., 2015; Russo, 2017). CGRP and amylin share ~50% sequence identity (Figure 1C), with a high proportion of amino acids being identical at the N and C terminus of amylin and αCGRP (Figure 1D). Should an antibody that was raised against one of these peptides recognise an epitope within these highly conserved regions, cross-reactivity with the other peptide is likely, with the likelihood of this increasing as peptide abundance increases. Therefore, investigations aiming to detect amylin in the nervous systems are potentially confounded by cross-reactive “amylin” antibodies that may be detecting the abundantly expressed CGRP, rather than amylin (H. Ghanizada, Al-Karagholi, et al., 2021). Similar considerations for investigations into CGRP expression in the pancreas should be made.

Although some studies have reported limited validation of anti-amylin antibodies against CGRP, such as determining cross-reactivity using immunoblotting or undertaking pre-adsorption controls, they are generally performed with CGRP concentrations likely to be insufficient to rule out the contribution of CGRP to amylin-like staining in neural tissues (Rees et al., 2021; Verma et al., 2016). Figure 3 describes potential outcomes from antibodies with different degrees of cross-reactivity. Our recent characterisation of a series of commercially available anti-amylin antibodies shows that this is more than just a theoretical concern and that amylin antibodies very frequently display cross-reactivity with CGRP (Rees et al., 2021). The remaining sections of this review discuss the evidence for peptide expression, but it must be taken in the light of the limitations of the tools and the degree of validation of those tools.

Figure 3: Cross-reactive anti-amylin and anti-CGRP antibodies can confound results.

A) The relationship between antigen expression and antibody cross-reactivity. Highly cross-reactive antibodies will display immunoreactivity in tissues with low to high antigen expression whereas, antibodies with low cross-reactivity will only display immunoreactivity if the tissue highly expresses the antigen. B) Cross-reactive anti-CGRP antibodies are known to stain the β-cells of pancreatic islets which highly express amylin and anti-CGRP antibodies that do not display cross-reactivity for amylin have been reported not to detect CGRP in pancreatic β-cells. C) Cross-reactive anti-amylin antibodies are known to stain neuronal vesicles which highly express CGRP and anti-amylin antibodies that do not display cross-reactivity for CGRP have been reported not to detect amylin in some neuronal vesicles. Created with BioRender.com.

α-CGRP and β-CGRP peptides

Overview and contribution of CGRP to migraine and pain

In humans, two isoforms of CGRP exist: α-CGRP and β-CGRP. α-CGRP is encoded by the CALCA gene, which also encodes pro-CT and CT and is alternatively spliced in a tissue-dependent manner (Amara et al., 1982). β-CGRP is encoded by a separate gene, CALCB (Amara et al., 1985). The two CGRP isoforms are both 37 amino acid neuropeptides, with the human isoforms sharing 92% sequence identity, differing by three amino acids (Bennett & Amara, 1992). Due to the similarities in their sequences the majority of studies do not distinguish between the isoforms and we will use CGRP as a collective term for both isoforms. However, the limited data suggest that α-CGRP and β-CGRP are expressed throughout the CNS and PNS where they display overlapping and discrete patterns of expression. Evidently, βCGRP is the predominant form in the enteric nervous system, with less α-CGRP expressed but this does not mean exclusive expression in this location, as is sometimes suggested (Mulderry et al., 1988; Russell et al., 2014; Schutz et al., 2004). Diverse biological functions have been attributed to CGRP including vasodilation, energy metabolism, cardiovascular homeostasis and nociception (Russell et al., 2014).

There is evidence for CGRP being involved in a wide range of pain disorders, including several forms of headache (Rosenfeld et al., 1983; Russell et al., 2014). The best characterised of these is migraine. Initial research showed that circulating CGRP concentrations are elevated during a migraine attack (Goadsby et al., 1990). Furthermore, the administration of exogenous CGRP is able to induce a migraine-like attack in migraine sufferers (Goadsby et al., 1990). The precise mechanisms employed by CGRP during migraine pathogenesis are unclear, although the trigeminovascular system plays a key role.

Increased CGRP immunoreactivity at the site of pain or in the associated nerve fibres and ganglia is also a common observation in several forms of chronic pain (Schou et al., 2017). This suggests that anti-migraine therapeutics that target the CGRP system could provide relief to chronic pain sufferers. For example, in arthritis, CGRP can accumulate in synovial fluid of the affected joint, activating the Aδ or C afferent sensory nerve fibres. During the development of arthritis these sensory nerve fibres from the DRG or TG become hyper-excitable, resulting in increased mechanical hyperalgesia (Bulling et al., 2001; Schaible et al., 2002). Studies using CGRP receptor antagonists and CGRP knockout mice suggest that the primary role of CGRP in arthritis is to centrally sensitise the dorsal spinal afferents to pain, with less CGRP activity associated with reduced pain behaviours and secondary hyperalgesia (Ebersberger et al., 2000; Hirsch et al., 2013). However, a clinical trial suggested that blocking CGRP action was not effective for reducing pain in patients with mild-moderate osteoarthritis (Jin et al., 2018). Although patients administered galcanezumab displayed reduced pain, when measured using OMERACT-OARSI criteria, this was not significantly different from placebo. The reasons for this are unclear but may relate to the disease state or low study retention and consequently low sample size.

The precise mechanisms linking CGRP to increased sensitivity to pain have not been fully described. However, the pain-sensing P2X₃ receptors may play an important role in both craniofacial and body pain (Aoki et al., 2004; Fabbretti et al., 2006). Several studies suggest that CGRP can directly or indirectly upregulate P2X₃ expression and trafficking to the cell surface (Dessem et al., 2010; Fabbretti et al., 2006; Simonetti et al., 2008). In addition to P2X₃, other cell surface proteins, such as the TRPV1 receptor and ASIC3 may also play important regulatory roles in CGRP action (Aoki et al., 2004; Dessem et al., 2010).

Where are α-CGRP and β-CGRP found in sites relevant to migraine and pain?

The detection of CGRP using mRNA-based in situ hybridisation or antibody-based immunoreactivity demonstrates good concordance, with approximately 30–50% of TG and DRG neuronal cell bodies from predominantly small-medium sized neurons expressing CGRP (Table 1). However, mRNA data suggest that CGRP is more highly expressed in the DRG than TG, with a 2.5–3-fold difference observed (Lopes et al., 2017). In TG and DRG, CGRP-like immunoreactivity (CGRP-LI) is typically observed as dense cytoplasmic staining in small neurons or granular staining in the vesicles of medium-large neurons. Pearl-like expression of CGRP in Aδ and C fibres is also commonly observed (Messlinger et al., 2020). CGRP expression has been detected across the three trigeminal nerve branches but potential differences in abundance or cellular localisation have not been detailed (Tornwall et al., 1996). Interestingly, in larger neuronal cell bodies, where a CGRP and AM receptor component, CLR, has been localised, CGRP does not appear to be expressed. This suggests that CGRP could act locally on nearby CGRP receptor expressing neurons, although possible autocrine mechanisms have been reported (Guo et al., 2020; Messlinger et al., 2020). The expression of CGRP in TG and DRG glial cells appears to be conflicting, with CGRP mRNA and immunoreactivity reported to be present or absent in different studies (Frederiksen et al., 2018; Lopes et al., 2017; Tajti et al., 2011). Definitively determining which isoforms of CGRP are expressed is complicated as antibodies are unlikely to distinguish between α-CGRP and β-CGRP. Data from α-CGRP knockout mouse models suggests that α-CGRP and β-CGRP are typically co-expressed, although the relative amounts vary. Interestingly, β-CGRP is expressed primarily in small neuronal cell bodies in the DRG as puncta, indicating a presence in vesicles, at a lower frequency than α-CGRP. This is supported by mRNA which suggests β-CGRP expression is up to 10-fold lower than that of α-CGRP (Flegel et al., 2015; Manteniotis et al., 2013; Schutz et al., 2004).

Table 1:

mRNA and protein expression of the calcitonin family of peptides in the TG, DRG and spinal cord.

mRNA
Peptide	TG			DRG			Spinal Cord			References
	Rodent	Human/Primate	Other	Rodent	Human/Primate	Other	Rodent	Human/Primate	Other
α-CGRP	YCG	YCF	Y	YCG	Y	Y	Y	Y	Y	(Bhatt et al., 2014; Edvinsson et al., 1998; Gibson et al., 1984; Lopes et al., 2017; Manteniotis et al., 2013; Ray et al., 2018; Tajti et al., 1999)
β-CGRP	Y	Y	-	YC	Y	Y	YC	W	Y	(Costigan et al., 2002; Flegel et al., 2015; Manteniotis et al., 2013; Noguchi et al., 1990a, 1990b; Ray et al., 2018; Toribio et al., 2003)
Amylin	Y1, C	W/N	-	Y1,C	W/N	-	W	W	N	(Costigan et al., 2002; Ferrier et al., 1989; Flegel et al., 2015; Isensee et al., 2014; Kogelman et al., 2017; Manteniotis et al., 2013; Martinez-Alvarez et al., 2008; Mulder et al., 1995; Mulder et al., 1997a, 1997b; Nicholl et al., 1992; Ray et al., 2018)
AM	Y	Y	-	YC	Y	-	YCF	Y	W	(Chen et al., 2016; L. Edvinsson et al., 2020; Fernandez et al., 2010; Flegel et al., 2015; Manteniotis et al., 2013; Martinez-Alvarez et al., 2008; Owji et al., 1996; Ray et al., 2018; Shan & Krukoff, 2001; D. Wang, Huo, et al., 2014; D. Wang, Li, et al., 2014; Zeng et al., 2014)
AM2	N	W/N	-	Y	W/N	-	W/N	W	W	(Flegel et al., 2015; Manteniotis et al., 2013; Martinez-Alvarez et al., 2008; Ray et al., 2018; Xiong et al., 2015)
Calcitonin	CG	-	-	-	-	N	-	-	N	(Park et al., 2011; Raddant & Russo, 2014; Rosenfeld et al., 1983; Toribio et al., 2003)
Pro-calcitonin	YG	-	-	-	-	-	-	-	-	(Park et al., 2011; Raddant & Russo, 2014; Tajti et al., 2011)
Protein
Peptide	TG			DRG			Spinal Cord			References
	Rodent	Human/Primate	Other	Rodent	Human/Primate	Other	Rodent	Human/Primate	Other
α-CGRP	YCF	YCF	YCF	YCF	YCF	YCF	YCF	YCF	YCF	(Cottrell et al., 2005; Dun et al., 1996; Edvinsson et al., 1998; Eftekhari & Edvinsson, 2011; Eftekhari et al., 2010; Frederiksen et al., 2018; Gibson et al., 1984; Hong et al., 2009; Hou et al., 2001; Patil et al., 2010; Tajti et al., 1999; Uddman et al., 2002; L. Zhang et al., 2001)
β-CGRP	-	-	-	YC	-	-	YCF	Y	-	(Kimura et al., 1987; Mulderry et al., 1988; Noguchi et al., 1990a, 1990b; Petermann et al., 1987; Schutz et al., 2004)
Amylin	C2,C	N	WCF	YC	-	-	YCF	-	-	(Edvinsson et al., 2001; L. Edvinsson et al., 2020; Gebre-Medhin et al., 1998; H. Ghanizada, Al-Karagholi, et al., 2021; X. Huang et al., 2010; Mulder et al., 1995; Mulder et al., 1997a, 1997b)
AM	YFGV	-	-	YCF	-	-	YCF	-	YCF	(Chen et al., 2016; L. Edvinsson et al., 2020; Fernandez et al., 2010; Hobara et al., 2004; Hong et al., 2009; Hong et al., 2010; Ma et al., 2006; Munoz et al., 2001; Owji et al., 1996; Serrano et al., 2002; D. Wang, Huo, et al., 2014)
AM2	-	-	-	YC	-	-	YF	-	-	(Cottrell et al., 2005; Xiong et al., 2015)
Calcitonin	C3,CF	-	-	-	-	-	-	-	-	(L. Edvinsson et al., 2020)
Pro-calcitonin	YG	-	-	-	-	-	-	-	-	(L. Edvinsson et al., 2020; Park et al., 2011; Raddant & Russo, 2014; Tajti et al., 2011)

mRNA expression as detected by PCR or in situ hybridization. Protein expression/immunoreactivity using antibodies or chromatography. Y= detected; W = weakly detected; N= not detected; C= conflicting information published; - = not done.

^CCell bodies of neurons;

^Ffibres;

^Gglia;

^Vvasculature.

¹Detected in rats but not mice.

²Antibody used is cross-reactive with CGRP.

³Antibody used was raised against CGRP not calcitonin.

CGRP mRNA and pearl-like immunoreactivity are prevalent in the fibres of the STN in both rodents and humans (Bhatt et al., 2014; Eftekhari & Edvinsson, 2011; Kogelman et al., 2018). The spinal cord is innervated by a dense network of CGRP immunoreactive fibres in laminae I and II of the dorsal horn which includes fibres originating in the TG and DRG (Table 1). Some CGRP immunoreactive fibres are also seen in the posterior horn and large motor neurons in the ventral horn. This is observed across species at cervical, lumbar, and thoracic levels of the spinal cord. CGRP also co-localises with other pain-related neuropeptides, such as substance P and pituitary adenylate-cyclase-activating polypeptide in synaptic vesicles (Eftekhari & Edvinsson, 2011; Gibson et al., 1984; Rosenfeld et al., 1983). β-CGRP mRNA and protein has been detected in in the superficial layers of the dorsal horn of α-CGRP knockout mice and human spinal cord, suggesting that β-CGRP may also be involved in nociception (Petermann et al., 1987; Ray et al., 2018; Schutz et al., 2004).

Differences in CGRP expression between males and females has been reported. Female mice had significantly fewer CGRP-LI neurons in C4, L3 and S2 DRG, compared to males (Yang et al., 1998). This difference appeared to be linked to the activation of the oestrogen receptor, which is co-expressed in DRG neurons. Ovariectomized female mice displayed a similar expression of CGRP to males. However, this expression was normalised when ovariectomized mice were treated with oestrogen, mimicking the lower expression observed in intact females (Yang et al., 1998). No sex differences were observed for CGRP mRNA in the TG of rats. However, in the medulla of female rats, CGRP mRNA was significantly greater than that in males (Stucky et al., 2011). Despite the high prevalence of migraine in females compared to males there are very few studies investigating differences in the expression of CGRP or its receptors between the sexes. Additionally, studies which do look at sex differences are primarily in animal models, therefore how this translates to humans is unknown.

CGRP concentrations are reportedly elevated in the plasma, synovial fluid and cerebrospinal fluid across a range of painful conditions; from tumour-associated pain to chronic lower back pain (Schou et al., 2017). This is consistent with migraine patients, where elevated plasma CGRP-LI has been observed during acute migraine attacks and in chronic migraine patients compared to healthy controls (Goadsby et al., 1990; Irimia et al., 2020). Interestingly, this elevation has been reported for up to 3 days following an acute migraine attack, suggesting sustained CGRP transcription and synthesis and/or release (Durham, 2006). Although elevated, the source is unclear. Triptans, which were until recently the only specific migraine treatments, were found to suppress CGRP promoter activity (Durham, 2006). Changes in CGRP expression have also been reported in response to opioids. Morphine treatment increased CGRP immunoreactivity in the DRG and spinal cord of rats, which appears to be mediated via another member of the CT peptide family, AM (Li et al., 2020). Administration of AM receptor antagonists in rats attenuated the morphine-dependent upregulation of CGRP-LI in DRG and spinal cord (D. Wang, Li, et al., 2014). In contrast, treatment with exogenous AM alone promoted CGRP expression (Hong et al., 2010).

Overall, these findings suggest that the expression of CGRP is a key factor in pain. CGRP release, activity or expression is consistently shown to be elevated in pain conditions. Several therapeutics have already been developed to block CGRP activity and so exploring their use in a broader range of painful conditions beyond migraine should be an active area of research. Furthermore, there are few drugs which lower CGRP expression and reduce its release. This is an area that can be further exploited.

Calcitonin

Overview and contribution of calcitonin and pro-calcitonin to migraine and pain

CT is a 32 amino acid peptide which is derived by proteolytic cleavage from pro-CT, a 116 amino acid pro-peptide. The major biological function of CT is calcium homeostasis via osteoclast-mediated reabsorption and kidney excretion (Chesnut et al., 2008; Warshawsky et al., 1980). Salmon CT (sCT), which shares only 50% sequence homology with human CT (hCT), is an approved drug for the treatment of hypercalcaemia, osteoporosis and other disorders (Chesnut et al., 2008). sCT is often used as a tool to investigate the actions of CT but interpretation of the data is complicated by its unique pharmacology. sCT is a potent and relatively long-acting agonist of both CT and AMY receptors (Andreassen et al., 2014; Poyner et al., 2002). This means that just because sCT may produce an effect, it does not mean that this is necessarily CTR-mediated as the effect could equally be occurring through an AMY receptor.

The biological functions of pro-CT are unclear, but it is known to be elevated in the circulation in inflammatory disorders and during bacterial infections. Pro-CT is considered an indicator of infection, as is the mature form of CT although this is less studied (Hamade & Huang, 2020). Elevated circulating pro-CT in chronic migraine sufferers has been reported, with a significant increase in pro-CT serum levels during migraine attacks when compared to the levels in the period between attacks (Leira et al., 2018; Turan et al., 2011). Interestingly, one study suggested that pro-CT can activate the CGRP receptor. This requires confirmation, but could indicate that pro-CT plays a role in nociception (Sexton et al., 2008).

Several studies have examined the ability of sCT or the closely-related eel CT (eCT) to treat pain (Humble, 2011; Schwartz et al., 1996; Terashima et al., 2019). The majority of this research has used exogenous administration of sCT and the role of endogenous mammalian CT in pain is less well described. However, hCT has been reported to display analgesic properties similar to sCT, whereas porcine CT induced hyperalgesia (Giusti et al., 1985). Due to the scarcity of experimental data using human or rodent CT, the discussion will focus on sCT and eCT.

CT is reported to have analgesic properties in a diverse range of pain conditions, including bone pain, cancer, migraine, osteoarthritis and osteoporosis (Chesnut et al., 2008; Humble, 2011; Micieli et al., 1988; Schwartz et al., 1996). This is particularly interesting as several of these studies were performed in human patients and reported sustained reductions in reported pain. Clinical studies have examined the therapeutic potential of sCT treatment in migraine patients. sCT treatment reduced the severity and frequency of migraine attacks (Micieli et al., 1988). The mechanism by which this occurred is unclear. Confusingly, one study observed increased circulating β-endorphin, a pain-reliving endogenous opioid, but also adrenocorticotropic hormone and cortisol, which are considered migraine promoting (Ustdal et al., 1989). However, another study reported no change in circulating β-endorphin (Micieli et al., 1988). Additionally, in patients suffering from atypical facial pain, which is primarily mediated by the TG, no response to sCT was observed, compared to placebo (Schwartz et al., 1996). Administration of sCT was also found to have an analgesic effect for patients suffering from spinal cord injuries with a beneficial effect on hyperalgesia and allodynia (Humble, 2011).

Rodent models have been used to explore the analgesic properties of sCT. For example, pre- or co-administration of sCT reduced threshold in the nociceptive hot plate test and reduced the frequency of pain-related behaviours, such as licking (Clementi et al., 1984; Hamada et al., 2018; Rahimi et al., 2019; Sibilia et al., 2000; Yeh et al., 2016). The proposed mechanism is attenuation of central sensitisation or the modulation of pain via serotonergic pathways. Co-administration of serotonin receptor antagonists, such as methysergide, attenuated sCT’s increased latency in hot plate tests (Clementi et al., 1984). Interestingly, sCT administration is also associated with lower neuronal activation in spinal cord and STN, measured as c-fos expression, and increased expression of serotonin and other monoamines in the periaqueductal grey (Khoshdel et al., 2014; Kilinc et al., 2018; Rahimi et al., 2019; Yeh et al., 2016). In the brain, sCT is reported to decrease neural sodium-dependent serotonin transporter activity, thereby potentiating serotonin activity in the synapse, resulting in increased 5-HT₁ receptor activity (Yeh et al., 2016). Additionally, eCT administration decreased expression of pro-excitability genes in the DRG, particularly sodium and calcium voltage gated channels, further modulating central sensitization (Ito et al., 2012; Terashima et al., 2019). A study using a migraine rodent model showed that sCT inhibited the degranulation of mast cells in the dura mater, thereby reducing activation of the trigeminal nerve and the subsequent release of CGRP (Kilinc et al., 2018).

Where are calcitonin and pro-calcitonin found in sites relevant to migraine and pain?

There is very little information available describing the expression of CT and pro-CT in spinal cord, TG or DRG (Table 1). Reports of pro-CT or CT mRNA expression in the TG are conflicting, with its presence and absence observed (Raddant & Russo, 2014; Rosenfeld et al., 1983). This may reflect the fact that mRNA encoding pro-CT, CT and α-CGRP are generated by alternate splicing of the CALCA gene and initial transcripts will contain RNA sequences encoding all three peptides (Amara et al., 1982). In rodent TG, pro-CT immunoreactivity was primarily observed in the soma of satellite glia and Schwann cells. Furthermore, pro-CT was not co-expressed with β-tubulin III, a neuronal cell marker (L. Edvinsson et al., 2020; Raddant & Russo, 2014; Tajti et al., 2011). Expression of pro-CT mRNA in glia and pro-CT-LI is strongly aligned suggesting that it is likely to be expressed in the glia of rodent TG. However, whether pro-CT is further processed to CT in these cells is unknown (Park et al., 2011; Raddant & Russo, 2014; Tajti et al., 2011). Although neither pro-CT nor CT mRNA have been detected in TG neurons, CT-LI was apparently detected (L. Edvinsson et al., 2020). Here, CT-LI was observed as granular cytoplasmic staining in the soma of small to medium neurons, satellite glia and C and Aδ fibres. However, the antibody used in these studies appears to be raised against CGRP, rather than CT, and is no longer available (L. Edvinsson et al., 2020). Therefore, this staining could actually be CGRP. Studies examining the expression of pro-CT or CT in DRG or spinal cord have not been conducted in rodents or humans (Table 1). However, in one study equine pro-CT and CT mRNA were not detected in the DRG or spinal cord (Toribio et al., 2003).

The analgesic properties of sCT and eCT are promising for the treatment of migraine and other pain disorders. These peptides are consistently demonstrated to be anti-nociceptive, though the exact mechanism of action is not fully elucidated. However, there is little data available examining the expression of pro-CT and CT in migraine and pain relevant structures, additionally, what is available is somewhat conflicting, making conclusions difficult. Further investigation into these peptides is needed.

Amylin

Overview and contribution of amylin to migraine and pain

Amylin, also known as islet amyloid polypeptide, is a 37 amino acid neuroendocrine hormone, which is co-secreted with insulin from pancreatic β-cells in response to food intake. Amylin acts via the brain to promote meal-ending satiation and to lower blood glucose by decreasing glucagon secretion (Hay et al., 2015). These functions led to the development of the amylin mimetic pramlintide. Pramlintide is an approved drug for the treatment of insulin-requiring diabetes and has been in clinical trials as a treatment for obesity (Hollander et al., 2004).

The potential role of amylin in pain has been investigated in several studies with somewhat contradictory results, suggesting that amylin can have both pro- and anti-nociceptive effects. Although the role of amylin in pain disorders is not well understood, studies examining amylin in primary headache disorders suggest a potential pro-nociceptive role for this peptide. In patients suffering from chronic, and to a lesser extent episodic, migraine displayed an elevated concentration of circulating amylin-LI, measured by ELISA, relative to non-migraineurs (Irimia et al., 2020). This elevation was independent of an individual’s metabolic status and obesity. Recent evidence suggests that amylin could play a direct role in triggering a migraine attack. The infusion of the amylin mimetic pramlintide into migraine patients was able to induce migraine-like attacks (H. Ghanizada, Al-Karagholi, et al., 2021). Pramlintide induced headache in 88% of patients and a migraine-like attack in 41% of patients. Symptoms associated with migraine attacks, including nausea, photophobia and phonophobia, were also observed in response to pramlintide. Interestingly, the administration of pramlintide was not associated with an elevation in circulating CGRP and the peak plasma concentration of pramlintide was insufficient to activate the CGRP receptor, as pramlintide is >1000-fold less potent than CGRP at this receptor (H. Ghanizada, Al-Karagholi, et al., 2021). A comparably low potency has been reported for amylin at the both the human and rat CGRP receptors (Walker et al., 2015). This suggests that the ability of pramlintide to induce migraine attacks is likely independent of both CGRP and the CGRP receptor. The ability of pramlintide to induce head pain is further supported by the prescribing information for diabetes treatment which lists headache as a potential adverse effect, with an incidence of 13% in the pramlintide group compared to 7% in the placebo group (San Diego, 2005).

Studies examining the effect of amylin on body pain in rodent models suggest the relationship is complex. Some studies show that administration of amylin in mice reduces pain-related behaviours, such as reducing writhing and tail flicks, and increases the latency in the hot plate test. This reduction in pain behaviour with amylin was consistently attenuated when amylin antagonists, such as AC187 and rat Amy_8–37 were administrated (X. Huang et al., 2010; Khoshdel et al., 2019; Khoshdel et al., 2016; Sibilia et al., 2000). Co-administration of amylin with morphine resulted in reduced morphine tolerance and amylin alone demonstrated an analgesic action comparable to morphine in tail flick tests (Khoshdel et al., 2019; Khoshdel et al., 2016). Interestingly, intrathecal, but not intraperitoneal, injection of amylin was able to reduce the number of formalin-induced flinches in pain phase I and II. This suggests central amylin administration is required to attenuate acute neurogenic pain mediated by nociceptive C-fibres in phase I and inflammatory pain mediated by the peripheral tissues and the spinal cord in phase II (Khoshdel et al., 2016; Potes et al., 2016). Additionally, intrathecal and intraperitoneal injection of amylin decreased the number of c-fos neurons in laminae I-II of the dorsal horn of the spinal cord, indicating a reduced activation of the spinal nociceptive mechanisms (X. Huang et al., 2010; Khoshdel et al., 2014). Furthermore, amylin knockout mice had reduced phase II pain behaviours following formalin injection (Gebre-Medhin et al., 1998). In contrast, subcutaneous co-administration of amylin with formalin appeared to increase the number of paw jerks in the acute phase I pain, whereas pre-administration increased paw jerks and focused pain behaviour in phase II (Potes et al., 2016). Central administration of amylin via the lateral cerebral-ventricle had no effect on nociception, suggesting that amylin’s nociceptive action is unlikely to be mediated by the hypothalamus or closely-related structures (Bouali et al., 1995). Only limited studies have been performed examining amylin activity in animal models of craniofacial pain, making conclusions difficult. In one study, peri-orbital administration of amylin to mice did not induce mechanical allodynia, suggesting that amylin does not directly activate or sensitise trigeminal afferents for pain (De Logu et al., 2019). However, it is possible that amylin receptors are not present in the terminals of trigeminal afferent nerves and the effect of amylin on pain occurs at a site physically separated from the terminals. For example, amylin may require access to neuron cell-bodies in the TG to contribute to craniofacial pain.

Where is amylin found in sites relevant to migraine and pain?

Relatively low levels of amylin mRNA are reported in the rat sensory ganglia, with higher levels detected in the DRG than the TG (Kogelman et al., 2017). While amylin mRNA appears to be consistently detected in rat, only very weak or no detection is observed in the sensory ganglia from mice and humans (Flegel et al., 2015; Kogelman et al., 2017; Manteniotis et al., 2013). In situ hybridization studies in rat indicates that amylin mRNA is expressed in the soma of approximately 15–30% neurons, predominantly of small to medium size (Table 1). Interestingly, some mRNA probes for amylin are reported to weakly detect CGRP and in this study, amylin-positive neurons represented a subpopulation of CGRP-positive neurons (Mulder et al., 1995). This is mirrored by immunoreactivity studies in rat TG and DRG where amylin-LI granular staining strongly co-localises with CGRP (L. Edvinsson et al., 2020; Gebre-Medhin et al., 1998; Mulder et al., 1995). Similar findings have able been reported in cat TG, with amylin-LI fibres also observed (Edvinsson et al., 2001). However, the antibody used to investigate amylin immunoreactivity in rat TG was later found to be cross-reactive with CGRP (L. Edvinsson et al., 2020; Rees et al., 2021). Additionally, in a separate study, amylin-LI in rat TG detected using anti-amylin antibodies which cross-react with rat CGRP colocalised strongly with CGRP immunoreactivity within puncta. However, when an anti-amylin antibody which does not cross-react with human CGRP was used, no observable amylin-like immunoreactivity was detected in human TG. This suggests that amylin-LI observed in the TG is likely due to cross-reactivity with CGRP and any observed staining likely correlates with the anti-amylin antibody’s cross-reactivity profile for CGRP (H. Ghanizada, Al-Karagholi, et al., 2021). Amylin-LI was abolished in the DRG of amylin knockout mice (Gebre-Medhin et al., 1998). Amylin mRNA was also strongly associated with TRPV1 and other molecules involved in pain in rat DRG neurons, consistent with a role in pain modulation (Isensee et al., 2014). Additionally, changes in amylin expression have also been linked to pain states. In DRG neurons, amylin mRNA was increased after exposure to noxious stimuli and down-regulated following spinal nerve ligation (Gebre-Medhin et al., 1998; Mulder et al., 1997b; Potes et al., 2016; H. Wang et al., 2002).

Amylin immunoreactive fibres and neurons were also reported in rat STN, the mesencephalic nucleus of the trigeminal nerve and the trigeminal root (D’Este et al., 2000; Skofitsch et al., 1995). These studies used antibodies which were pre-adsorbed with relatively low concentrations of rat and human CGRP compared to the concentrations in a vesicle (Russo, 2017). Additionally, pre-adsorption with rat and human amylin failed to completely block staining in these structures, indicating the potential for some cross-reactivity. In spinal cord, low levels of amylin mRNA have been detected across species. Rodent immunoreactivity studies suggest that amylin may be present in fibres of superficial levels of the dorsal and ventral horn, strongly co-localising with CGRP. Amylin-LI fibres appeared to be distributed across all levels of the spinal cord and loss of staining was also observed in amylin knockout mice (Gebre-Medhin et al., 1998; X. Huang et al., 2010; Mulder et al., 1995; Mulder et al., 1997a) (Table 1). While the absence of immunostaining in the DRG and spinal cord of amylin knockout mice is compelling it is difficult to conclude that amylin is present in these tissues, particularly as studies have reported very low or no amylin mRNA in mice and human. Further investigation using well validated amylin antibodies should give greater insight into whether amylin is present in these tissues.

Collectively, these findings indicate that amylin plays a role in sensory transmission and nociception in the DRG and spinal cord but its expression in these structures, or the TG, is not conclusive. Overall, the ability of amylin administration to modulate pain behaviour and downstream neural activation appears to be dependent of the site of administration, the duration of amylin exposure and whether is administered before or co-administered with the painful insult. This demonstrates that amylin’s contribution to nociception is complicated and requires further investigation (Almeida et al., 2019; Bouali et al., 1995; Potes et al., 2016). It is possible that different AMY receptor subtypes could be responsible for these effects, or if a very high dose of amylin is used, it could be cross-reacting with a different CT family receptor. Where the actions of amylin and CGRP appear to be identical in a given model, this could point to the involvement of a receptor that is shared between both peptides.

Adrenomedullin and adrenomedullin-2

Overview and contribution of adrenomedullin and adrenomedullin-2 to migraine and pain

AM is a 52 amino acid peptide which is primarily expressed by endothelial cells. AM is a potent vasodilator, has cardio-protective effects and plays important roles in angiogenesis and lymphangiogenesis (S. Y. Zhang et al., 2018). The closely related, AM2, is a 53 amino acid peptide which is also expressed by endothelial cells. The biological actions of AM2 are not described in detail, however, it appears to have many overlapping functions with AM (Hong et al., 2012; S. Y. Zhang et al., 2018). Distinct activities of AM2 have also been reported in specific brain regions and other tissues (Hong et al., 2012). The mature forms of AM and AM2 have an extended N-terminus compared to other members of the calcitonin peptide family. These N-terminal residues do not appear to be required for activity (Hay et al., 2018; Hong et al., 2012). Interestingly, circulating pro-AM and mature AM levels are raised in several painful and inflammatory conditions, such as acute appendicitis, systemic inflammatory response syndrome and acute myocardial infarction (Haaf et al., 2013; Miguez et al., 2016; Ueda et al., 1999). Understanding the role of AM is complicated as relatively high concentrations, which can mimic CGRP and activate the CGRP receptor (Figure 2), are routinely used in research (Hay et al., 2018). However, several lines of evidence suggest there is a potential role for AM in pain. Whether AM2 is involved in pain is not known, however the overlapping activities reported with AM it would be unsurprising if a role in pain was identified.

The ability of AM to provoke a migraine-like attack in migraine patients has been examined in two clinical studies. In the first study, AM infusion induced facial flushing and superficial temporal artery dilatation. However, there was no associated increase in cerebral blood flow which had previously been observed in CGRP-induced migraine-like attacks (Petersen et al., 2009). Although a tendency towards increased headache was observed for AM infusion (58% of participants), this was not significantly different from the placebo group (33% of participants) up to 24 hours post infusion. This suggested that AM was unable to provoke a migraine-like attack or headache in individuals who suffer from migraine without aura (Petersen et al., 2009). However, it is possible that an effect could have been masked due to the small number of patients and the relatively high placebo response (H. Ghanizada, Iljazi, et al., 2021).

In a more recent study, AM infusion was reported to significantly provoke migraine-like attacks (55% of participants) compared to placebo (15% of participants) in migraine without aura patients. Furthermore, the headache intensity scores and the overall incidence of headache (80% of participants) were significantly higher following AM infusion than placebo (15% of participants) (H. Ghanizada, Al-Karagholi, M.A., Arngrim, N., Mørch-Rasmussen, M.K., Walker, C.S., Hay, D.L, Ashina, M., 2021). Additionally, AM induced cardiovascular effects including increased heart rate and facial flushing. The different conclusions in the two studies are somewhat unexpected as both had similar experimental design. However, small differences in the amount of intravenous AM infused (0.08 μg/kg/min vs. 0.12 μg/kg/min for 20 minutes) and the number of migraine without aura patients studied (12 vs. 20 patients) likely resulted in the second study having greater statistical power. Interestingly, AM-induced headache was reported as a side-effect in ~75% of individuals with hypertension in a clinical study examining the haemodynamic and urinary effects of AM (Troughton et al., 2000). It is important to note that whether AM-induced migraine-like attacks or headache is due to activation of an AM receptor or the off-target activation of the CGRP receptor is not clear. The effect of AM2 has not been investigated in migraine or headache clinical trials.

AM is reported to have pro-nociceptive functions and may play a role in a diverse range of different pain types (Chen et al., 2016; Hong et al., 2010; D. Wang et al., 2013). The intrathecal administration of AM can induce thermal hyperalgesia and mechanical allodynia, suggesting that AM can mediate pain and pain sensitisation (Hong et al., 2009; Ma et al., 2006; Sugimoto et al., 2013). Interestingly, mechanical allodynia was reportedly greater in rats that were administered repeated doses of AM, highlighting the potential contribution to pain sensitisation (H. Huang et al., 2019). The effects of AM on pain have been further investigated though the use of the AM receptor antagonist AM_22–52. For example, the effects of AM on mechanical allodynia could be abolished when co-administered with the AM_22–52 (H. Huang et al., 2019). Several lines of evidence suggest that endogenous AM can contribute to acute inflammatory pain. In rodent models, pro-inflammatory treatments, such as Freund’s complete adjuvant, formalin and capsaicin, were associated with an increase in AM expression in the DRG and spinal cord (Hong et al., 2009; Ma et al., 2006; Sugimoto et al., 2013). Hyperalgesia induced by these pro-inflammatory treatments and formalin induced phase II flinching are reportedly attenuated by AM_22–52, suggesting that the upregulated AM was contributing to pain via AM receptors. Additionally, a nervous system specific AM knockout mouse model showed prolonged tail flick latency, a test of spinal reflexes, and a shortened latency in the hot plate test (Fernandez et al., 2010). Together, this suggests that AM may contribute to nociception in the spinal cord, but may have an anti-nociceptive role through supraspinal mechanisms.

AM is reported to modulate thermal and mechanical hyperalgesia through a number of signalling molecules and pathways. Exogenous AM is reported to elevate the expression of c-fos, CGRP, TRPV1, NO and a number of inflammatory cytokines in the DRG and spinal cord (Li et al., 2020). Additionally, AM is reported to promote changes in not just the neurons, but also glia of the DRG and spinal cord (Zeng et al., 2014). For example, AM promotes morphological changes in glia, such as hypertrophy, and increases the expression of inflammatory molecules. AM_22–52 can attenuate these responses in rat DRG and spinal cord explants (Zeng et al., 2014). Understanding how AM modulates glia is of particular importance as glial activation plays a critical role in spinal processing of pain signalling. AM also appears to be involved in morphine tolerance, as exogenous AM exacerbates tolerance and AM_22–52 decreases tolerance (Li et al., 2020; D. Wang, Huo, et al., 2014). AM appears to mediate morphine tolerance via mechanisms similar to heat hyperalgesia and mechanical allodynia (Hong et al., 2010; Li et al., 2020; Zeng et al., 2014).

Where are adrenomedullin and adrenomedullin-2 found in sites relevant to migraine and pain?

AM and AM2 display widespread expression throughout the body, likely due to their expression in vascular endothelial cells (Karpinich et al., 2011; Morimoto et al., 2007). However, few studies have investigated the expression of AM in the TG. AM mRNA has been detected in the TG of rodents and humans (Flegel et al., 2015; Manteniotis et al., 2013) (Table 1). Consistent with this finding, AM-LI has been observed in blood vessel walls associated with the TG, indicating vascular endothelial staining, and in the cytoplasm of satellite glia and Schwann cells (L. Edvinsson et al., 2020; Flegel et al., 2015; Manteniotis et al., 2013). AM-LI has also been reported to co-localise with CGRP in the soma of small to medium neurons in the DRG and large neurons within the STN (Ma et al., 2006; Macchi et al., 2006; Munoz et al., 2001). AM2 mRNA is not detectable in the TG and currently no studies have investigated AM2 protein expression (Flegel et al., 2015; Manteniotis et al., 2013).

AM, and relatively low levels of AM2, are reported to be expressed in the DRG, with a consensus between mRNA and immunoreactivity studies (Flegel et al., 2015; Manteniotis et al., 2013; Ray et al., 2018) (Table 1). In rodents, AM-LI and AM2-LI are present in a subset of small to medium neuronal cell bodies; whether AM and AM2 are expressed within the same neurons is yet to be determined. Diffuse AM staining, which strongly co-localises with CGRP and TNF-α is observed in approximately 30% of neuronal cell bodies (Chen et al., 2016; Cottrell et al., 2005; Fernandez et al., 2010; Hobara et al., 2004; Hong et al., 2009; Ma et al., 2006). Additionally, AM-LI in the DRG is abolished in AM knockout mice (Fernandez et al., 2010). Interestingly, DRG neurons expressing AM also co-express AM receptor components, CLR, RAMP2 and RAMP3, suggesting that AM may act in an autocrine manner (Hong et al., 2010).

AM expression has been reported in the STN. AM mRNA is moderately expressed in the STN at similar levels to that in spinal cord (Kogelman et al., 2018). AM-LI in large neurons and fibres in the STN and tract have been reported (Macchi et al., 2006; Serrano et al., 2002). Staining was reduced when pre-adsorbed with AM but not CGRP, suggesting that immunoreactivity was due to genuine AM detection, rather than antibody cross-reactivity with abundant CGRP (Macchi et al., 2006). In the spinal cord, mRNA analysis using in situ hybridisation and immunoreactivity suggest that AM is primarily present in the soma of small to medium neurons and fibres in laminae I and II of the dorsal horn at the cervical, thoracic, and lumbar levels (Table 1). Furthermore, some AM-LI was observed in the ventral horn and in large neurons of the anterior horn (Hong et al., 2009; Hong et al., 2010; Ma et al., 2006; Serrano et al., 2002; Shan & Krukoff, 2001). Interestingly, although AM was co-expressed with CGRP in IB4 (C-fibre marker) expressing neurons, there was little co-localisation, suggesting that AM and CGRP are localised in different subpopulations of vesicles (Ma et al., 2006). AM2-LI fibres and mRNA expression has also been observed in the dorsal horn (Cottrell et al., 2005). Investigation into the expression of AM and AM2 has focussed on rat spinal cord; whether these findings translate to humans is unknown. AM-LI is reported to be upregulated in the DRG and spinal cord in response to noxious stimuli (Hong et al., 2009). Additionally, changes in AM mRNA and immunoreactivity have been observed in DRG and spinal cord in response to pain related treatments, particularly morphine administration (Hong et al., 2010; D. Wang, Li, et al., 2014).

The available AM mRNA and immunoreactivity data within structures related to migraine and pain are in consensus across species, suggesting that it is likely expressed in these structures and may contribute to pain. AM has a potential pro-nociceptive role, which appears to be partially mediated by upregulating CGRP. AM and the AM receptors could provide additional avenues for targeting CGRP-based pathophysiology, or as targets in their own right, and should be considered for further investigation and potential clinical development. Additionally, it cannot be ruled out that high levels of AM are not activating the CGRP receptor.

Future directions

Members of the CT peptide family display both pro- and anti-nociceptive activities which could be clinically exploited. However, there is very limited understanding of the mechanisms behind these activities, with conflicting data reported. Several members of the CT peptide family appear to be expressed at multiple sites that could contribute to pain, but further investigation is required (Figure 4). Not only is there a lack of localisation data but the data that is available is hard to interpret due to poorly characterised or imprecise tools. Overall, there are major gaps in our knowledge of where these peptides are expressed and potentially contributing to pain transmission or sensitisation (Table 1). Future studies should focus on elucidating the specific expression of CT peptides family members using thoroughly validated mRNA probes and antibodies. Alternatively, techniques which avoid the use of probes altogether, such as mass spectrometry imaging, should be strongly considered. The application of techniques, such as high-throughput single cell RNAseq, which provide higher resolution data, could allow a better molecular understanding of key subpopulations involved in nociception, possibly providing additional therapeutic targets (Ray et al., 2018). Translating findings from rodent and other animal models to humans should also be prioritised as some species differences have been observed.

Figure 4: Summary of the expression of the CT family of peptides in the TG, DRG, spinal cord and STN of rodent and humans.

Distribution of α-CGRP, β-CGRP, AM, AM2, Pro-CT, CT and amylin mRNA and protein in structures related to migraine and pain. Expression where reports are conflicting or weak/no expression is observed have been omitted. Interpretation of these data is difficult as the mRNA probes and in particular antibodies used have limited characterisation, therefore caution should be applied. In some cases, the expression of a particular peptide has not been investigated, leaving a knowledge deficit. Created with BioRender.com.

Conclusions and Implications

The recent approval of anti-migraine therapies targeting the CGRP system has dramatically improved the quality of life for many sufferers and may offer hope for many patients of other chronic pain disorders who have an unmet clinical need for pain relief. However, a subset of migraine patients do not respond well to these treatments, suggesting additional clinical targets or strategies are necessary. In this review, we have looked beyond CGRP and considered the role the other members of the clinically important CT peptide family play in nociception and where they are localised at sites relevant for migraine and other pain conditions. It is particularly important to consider all members of the family collectively, as these peptides display overlapping pharmacology, distribution and potentially function. Data interpretation for the entire family is complicated by the use of probes which are poorly validated or cross-reactive to detect these peptides. There is still more that needs to be learned. In conclusion the CT peptide family contains potential novel and untapped clinical targets for migraine and other painful conditions.

Abbreviations

CGRP

calcitonin gene-related peptide

CT

calcitonin

Amy

amylin

AM

adrenomedullin

AM2

adrenomedullin 2 or intermedin

TG

trigeminal ganglia

DRG

dorsal root ganglia

STN

spinal trigeminal nucleus

CRSP

calcitonin receptor-stimulating peptide

CLR

calcitonin receptor-like receptor

CTR

calcitonin receptor

RAMP

receptor activity-modifying protein

AM₁ receptor

adrenomedullin 1 receptor

AM₂ receptor

adrenomedullin 2 receptor

AMY₁ receptor

amylin 1 receptor

AMY₂ receptor

amylin 2 receptor

AMY₃ receptor

amylin 3 receptor

CNS

central nervous system

PNS

peripheral nervous system

sCT

salmon calcitonin

eCT

eel calcitonin

CALCA

calcitonin related polypeptide alpha

CALCB

calcitonin related polypeptide beta

P₂X₃ receptor

P2X purinoceptor 3

TRPV1

vanilloid receptor

ASIC3

acid-sensing ion channel 3

CFA

complete Freund’s adjuvant

IB4

isolectin B4

Data availability statement

Data sharing is not applicable to this article because no new data were created or analysed in this study.