1. Introduction
Migraines are a debilitating condition afflicting approximately 30% of the world population. Migraine pain is frequently accompanied by intracranial hypersensitivity to stimuli and central sensitization or allodynia [
1,
2]. Migraineurs frequently report cephalic and extracephalic allodynia, including hypersensitivity to stimuli and pain (hyperalgesia), and tenderness in the head and extremities [
3]. Advances in understanding the molecular pathways underlying hypersensitivity to pain have been gained from studying the expression profile of genes as a whole [
4]. This study of gene expression in the trigeminal ganglia and nucleus accumbens of mice presenting cephalic and extremity hypersensitivity to pain elicited by nitroglycerin (NTG) identified significant differences in 275 genes relative to the control. The previous genes were annotated to the circadian rhythm and glutamatergic and dopaminergic synapse processes, including prohormone genes in signaling pathways that code for neuropeptides such as proenkephalin (PENK) and natriuretic peptide type C (NPPC), and neuropeptide-like protein chromosome 4 open reading frame (C4orf48) [
4]. Reports of the role of neuropeptides such as pituitary adenylate-cyclase-activating polypeptide (PACAP) and receptors in migraine [
5,
6,
7] and retinal and ocular disorders [
5,
8,
9,
10,
11] emphasize the participation of these signaling molecules across central nervous system regions.
Despite the attention to gene pathways and profiles, less attention has been directed to understanding the relationship between hyperalgesia and the individual transcript isoforms resulting from alternative splicing. Evidence is accumulating about the relationship between alternative splicing products and migraine, pointing to new insights into the disease. For example, splice variation in calcium channel pore-forming α1-subunit gene has been associated with familial hemiplegic migraine [
12], and sodium voltage-gated channel alpha subunit 1 (SCN1A) has been associated with migraine without aura [
13]. Two recent reviews have highlighted the role of the calcitonin gene-related peptide (CGRP) on migraines and effective treatments [
14,
15]. The neuropeptide CGRP results from cleavage of a calcitonin-related polypeptide alpha (CALCA) isoform, but bioactive neuropeptides resulting from other CALCA isoforms including calcitonin are not known to participate in migraine. Two of the isoforms produced by the alternative splicing of the gene cocaine- and amphetamine-regulated transcript protein (CARTPT) are translated into bioactive peptides [
16] involved in pain signaling processing [
17].
Many neuropeptide prohormone genes such as CALCA, CARTPT, and tachykinin precursor 1 (TAC1) undergo alternative splicing, with the resulting neuropeptide products having different functions. The levels of alternative splicing isoforms from 38 neuropeptide prohormone and receptor genes were affected by maternal immune activation in pig hippocampus and amygdala [
18]. The impacted isoforms included peptide YY (PYY), calcitonin-related polypeptide beta (CALCB), relaxin 2 (RLN2), secretogranin II (SCG2), neuropeptide Y receptor Y5 (NPY5R), glucagon-like peptide 1 receptor (GLP1R), and oxidized low-density lipoprotein receptor 1 (OLR1) [
18]. The previous alternative splicing events may impact migraine and pain because maternal immune activation during gestation has been associated with autism and schizophrenia spectrum disorders that present a high incidence of migraine and pain sensitivity comorbidities [
19,
20].
Although there is ample evidence supporting the association between hypersensitivity to pain and neuropeptide signal–receptor systems, the participation of gene isoforms is limited. This situation is partly due to the small size of many neuropeptide isoforms, resulting in skewed detection distribution. Therefore, standard models have limited statistical power to test for differential abundance or splicing, even in experiments with adequate sequencing depth. Accurate testing of differential neuropeptide isoform levels necessitates non-standard zero-inflated models [
21,
22].
With this study, we aim to advance the understanding of changes in alternative splicing processes and events in neuropeptide prohormone and receptor genes associated with hypersensitivity to pain. Sequenced reads from an RNA-seq experiment that profiled two central nervous system regions from mice presenting hypersensitivity to pain and controls were used. The standard gene-centric expression study [
4] was superseded using de novo assembly and prediction methods to reconstruct all transcript isoforms in the samples. Zero-inflated models were used to identify neuropeptide prohormone and receptor transcript isoforms with differential expression due to hyperalgesia. Alternative splicing events of neuropeptide prohormone and receptor genes associated with hyperalgesia were identified. The combination of de novo assembly, differential transcript expression, and differential splicing events were used to identify alternative splicing of neuropeptide and receptor genes that underly hyperalgesia phenotypes.
4. Discussion
Insights into the participation of alternative splicing in neuropeptide prohormone and receptor genes on hypersensitivity to pain, a condition frequently experienced by migraineurs, were gained by advanced isoform analysis of transcriptome profiles [
4]. De novo assembly of the reads from an RNA-seq study of the TG and NAc enabled thorough detection of all the isoforms across samples. Detection was followed by testing for differential isoform levels between treatment and region groups using zero-inflation models and differential splicing events. The range of alternative splicing mechanisms detected in the present study offers insights into the variety of processes by which hyperalgesia conditions could regulate gene expression and neuropeptide or receptor production.
Differential alternative splicing events associated with hyperalgesia in the untranslated regions (UTRs) were detected in neuropeptide prohormone and receptor genes with different transcription start sites or non-coding exons. Alternative splice sites involving the UTR impact the stability, translational efficiency, and localization of the mRNA [
49], suggesting additional ways by which mRNA gene expression is modulated in hyperalgesia conditions. Differential splicing in the UTR of VGF, a single-exon coding gene detected in the present study, was supported by the de novo assembly recovery of the complete mRNA sequence in all 20 samples. Moreover, VGF transcript isoforms exhibited significantly higher expression in NAc than TG. The association between VGF splicing and hyperalgesia is consistent with reports that VGF tends to be overexpressed in pain signaling structures of many neuropathic pain animal models [
50]. Reports of elements that regulate the expression of VGF through action in the promoter and single exon could relate to the mutually excluded non-coding exon detected in the present study [
51].
Other types of differential alternative splicing events associated with hyperalgesia encompassed changes in the translated mRNA sequence resulting in new neuropeptides or the absence of known neuropeptides. The impact of nucleobindin 2 (NUCB2), SCG2, and SCG3 alternative splicing events identified by rMATS are unknown, but the SCG2 events would remove an experimentally uncharacterized neuropeptide. Neuropeptide prohormone genes TAC1, CALCA, and IGF1 and neuropeptide receptor genes ADCYAP1R1, CRHR2, and IGF1R presented alternative splicing events, and changes in the resulting isoforms may impact the presence and severity of hyperalgesia.
The alternative splicing of TAC1 resulted in four mutually exclusive transcript isoforms that were also identified in de novo assembly. While only two transcript isoforms are currently annotated in the mouse assembly, these transcript isoforms correspond to human and rat α-, β-, γ-, and δ-TAC1 isoforms [
52]. A schematic representation of the Tac1 gene (
Figure 1) illustrates the differences between transcript isoforms and peptide production of the neuropeptides neuropeptide K, neuropeptide gamma, and neurokinin A. The predominant alternative splicing events detected by rMATS were the skipping of exons 4 and 6. An additional rare event (<1% of assigned reads) corresponded to skipped exons 3 and 4. The inclusive level difference indicates a higher inclusion of exons 4 and 6 in NAc than in TG (
Figure 1), for example, a 0.6 inclusive level difference was observed between NAc and TG (denoted as na_tg in
Figure 1). While all four protein isoforms produce substance
p,
Figure 1 reveals that neuropeptide K is produced only by β-TAC1, neuropeptide gamma is only produced by γ-TAC1, and neurokinin A can be produced from both β-TAC1 and γ-TAC1. The α-TAC1 and δ-TAC1 isoforms have truncated peptide sequences of neuropeptide K and neuropeptide gamma, respectively, that lack the C-terminal neurokinin A sequence and associated cleavage site. Among the genes presenting mutually exclusive events in the NAc and TG, neurokinin A tends to be associated with migraine and pain [
53], while neuropeptide K tends to be associated with hypotension [
52].
The known alternative splicing events in CALCA correspond to mutually exclusive exons and result in different protein isoforms that contain the neuropeptides CGRP and calcitonin. While de novo assembly identified the CGRP-containing transcript isoform, a possible calcitonin transcript isoform was identified. This transcript included an intron that would provide the calcitonin transcript isoform when spliced. Transcripts for both protein isoforms and the CALCR were underexpressed in NTG compared to CON, and the calcitonin producing isoform was underexpressed in NAc compared to TG. The significant alternative splicing of CALCA between NTG and CON detected by rMATS corresponded to alternative 3′ and 5′ splice site events in the 5′ untranslated region of a non-coding exon. Aligned with our finding of differential splicing in the 5′ untranslated region of CALCA, a single nucleotide polymorphism (rs3781719) located in the promoter region of CALCA has been associated with the response of patients to the OnabotulinumtoxinA chronic migraine therapy [
54].
IGF1 has 20 transcript isoforms arising from different alternative start sites and alternative terminal regions that can be translated into 15 unique protein sequences. Differential expression between hyperalgesia and control groups was identified in the IGF1 isoforms encompassing the terminal region. The association between IGF1 isoforms and hyperalgesia detected in this study is supported by reports that the administration of IGF1 is proven effective in alleviating sensory neuropathy [
55]. Notably, more unique IGF1 protein sequences (NP_001104745.1 and NP_001300939.1) were encoded by the isoforms in the TG than in the NAc. Associations between spliced products of the IGF1 and sensitivity to stimuli comorbidities have been reported. An isoform of IGF1 was differentially expressed in the hippocampus between pigs exposed to maternal immune activation during gestation and controls [
18]. Maternal immune activation has been associated with behavioral disorders, including autism and schizophrenia spectrum disorders and sensitivity to stimuli [
20].
Isoforms coded by the receptor ADCYAP1R1 exhibited significant differential expression and alternative splicing events between the hyperalgesia and control groups. Differential expression of ADCYAP1R1 transcript isoforms in the pig hippocampus was associated with maternal immune activation [
18]. Two almost identical skipped exon events in ADCYAP1R1 corresponded to the hip and hop exons [
56,
57], and the differentially expressed isoform lacked the hop exon. The association between nitroglycerin-induced hypersensitivity to pain and splicing events detected in the present study could be related to the requirement of the pituitary adenylate cyclase-activating polypeptide for the development of spinal sensitization and establishment of neuropathic pain [
58]. The activity of ADCYAP1R1 can be fine-tuned through alternative splicing variants that modify the ligand binding and signaling properties of this receptor [
57].
All de novo predicted neuropeptide prohormone gene isoforms were consistent with corresponding peptides experimentally identified in 10 other neuropeptidomic studies. The de novo assembly presented in this study enabled the detection of 250 isoforms from 75 neuropeptide genes and 323 isoforms from 83 receptor genes. In addition, de novo predicted previously undiscovered transcript isoforms in AUGN, chromogranin A (CHGA), and TAC1. The novel AUGN isoform involved an alternative 3′ splice site and would result in a truncated augin neuropeptide. The novel CHGA isoform involved a skipped exon and would result in a truncated beta-granin neuropeptide. This information can be used in neuropeptidomic studies to find novel peptides; for example, information from de novo transcripts was used to detect neuropeptides in nudipleuran gastropods by mass spectrometry [
59].
Differential alternative 3′ splice site splicing events in CRHR2 between hyperalgesia and control groups were observed in the TG and across regions. The CRHR2 splicing event involves modifications in the α1 exon comparable with the products of different initiation sites between CRHR2 isoforms [
60]. The changes in the alternative splicing of CRHR2 could be related to abnormalities in CRHR2 correlated with pain and the use of CRHR2 antagonists against chronic pain [
61]. Differential alternative 3′ splice site splice events were detected in IGF1R between the hyperalgesia and control groups within the NAc. The splicing events in IGF1R associated with hyperalgesia could correspond to the effect of IGF1R inhibition on reducing thermal and mechanical pain hypersensitivity [
62]. Altogether, these results highlight a range of alternative splicing and isoform expression processes that can modulate the levels of neuropeptide prohormone and receptor mRNA that participate in hypersensitivity to pain.
The significant differences between hyperalgesia and control mice detected for isoforms in other genes are aligned with reports of the association between these genes and pain-signaling phenotypes. Both the neuropeptide prohormone gene and an associated receptor gene were differentially expressed. Opioid peptides from PENK and prepronociceptin (PNOC) and opioid receptor delta 1 (OPRD1) are associated with pain [
63,
64]. Neuropeptide FF-amide peptide precursor (NPFF) is associated with migraine, and the NPFF receptor, neuropeptide FF receptor 2 (NPFFR2), has a role in the modulation of nociception [
65]. NPPC and an associated receptor, natriuretic peptide receptor 3 (NPR3), are associated with inflammation [
66].
Neuropeptides directly associated with pain or migraine from significantly differentially expressed neuropeptide prohormone genes include apelin (APLN) [
67,
68], cortistatin (CORT) [
69], IGF2 [
70,
71], neuromedin B (NMB) [
72], neuropeptide W (NPW) [
73], PDGFA and platelet-derived growth factor, D polypeptide (PDGFD) [
74], TAC4 [
75], and VIP [
76]. The role of some neuropeptide prohormone genes may be indirectly associated with migraines or pain. Pro-melanin concentrating hormone (PMCH) has been associated with the influence of eating and glucose on migraines [
77]. Natriuretic peptide type B (NPPB) and natriuretic peptide type C (NPPC) [
78] and SCG3 [
79] may act via the cardiovascular and neurovascular regulation.
Receptors directly associated with migraine or pain include endothelin receptor type B (EDNRB) [
80], GIPR [
81], and hypocretin receptor 1 (HCRTR1) [
82]. Other receptors with possible indirect effects on migraine are parathyroid hormone 1 receptor (PTH1R) and PTH2R, which mediates calcium and phosphate homeostasis [
83], and NPY5R that acts as receptor for neuropeptide Y (NPY), PYY, and PYY, and influences eating [
84,
85].