Glutamatergic Systems in Neuropathic Pain and Emerging Non-opioid Therapies

Abstract

Neuropathic pain, a disease of the somatosensory nervous system, afflicts many individuals and adequate management with current pharmacotherapies remains elusive. The glutamatergic system of neurons, receptors and transporters are intimately involved in pain but, to date, there have been few drugs developed that therapeutically modulate this system. Glutamate transporters, or excitatory amino acid transporters (EAATs), remove excess glutamate around pain transmitting neurons to decrease nociception suggesting that the modulation of glutamate transporters may represent a novel approach to the treatment of pain. This review highlights and summarizes (1) the physiology of the glutamatergic system in neuropathic pain, (2) the preclinical evidence for dysregulation of glutamate transport in animal pain models, and (3) emerging novel therapies that modulate glutamate transporters. Successful drug discovery requires continuous focus on basic and translational methods to fully elucidate the etiologies of this disease to enable the development of targeted therapies. Increasing the efficacy of astrocytic EAATs may serve as a new way to successfully treat those suffering from this devastating disease.

Graphical Abstract

1. Introduction to Neuropathic Pain

Pain is an unpleasant sensory experience that results from actual or perceived tissue damage [1]. Often this physiologic response to noxious stimuli or disease may be protective, as it prompts us to seek care or prevent further damage. However, when pain lasts beyond the acute period of injury, or longer than three months, it becomes a chronic disease [2]. Chronic pain is a common reason for seeking medical care and is associated with an increased incidence of poor mental health, opioid dependence, and a reduced quality-of-life [1, 3, 4].

Various types of chronic pain exist with unique characteristics and pathologies, including nociceptive pain, neuropathic pain (NP), inflammatory pain, nociplastic pain, and mixed pain, among others (Figure 1). The International Association for the Study of Pain (IASP) classifies chronic pain conditions into several categories for the purposes of the International Classification of Diseases (ICD-11) [2]. The IASP proposes two broad categories: chronic primary pain and chronic secondary pain, or pain secondary to an underlying disease. Chronic primary pain cannot be accounted for by another condition, occurs in one or more anatomical regions, and is associated with significant disability or distress. This includes fibromyalgia (chronic widespread pain), complex regional pain syndrome and nonspecific low-back pain. Chronic secondary pain syndromes occur secondarily to other diseases. These pathologies are categorized into subgroups: chronic cancer-related pain, chronic secondary visceral pain, chronic NP, chronic secondary headache and orofacial pain, chronic posttraumatic and postsurgical pain, and chronic musculoskeletal pain. However, chronic headache and orofacial pain, visceral pain and musculoskeletal pain could also be classified as chronic primary pain conditions if the etiology cannot be tied to another primary disease [2]. Despite a systematic classification system for chronic pain conditions, there remains an ongoing debate on how multifactorial pain disorders are categorized.

Figure 1. Overview of Pain Types.

There is overlap between pain types, however, pain disorders are often grouped into several broad categories: Nociceptive, inflammatory, neuropathic, nociplastic, dysfunctional/mixed pain. A short description of each type is provided in the middle text box (please refer to text for more information) along with examples of diseases in each category in the bottom text box.

Created with BioRender.com

The prevalence of chronic pain is high among adults in the United States and worldwide. The National Center for Health Statistics report from the Centers for Disease Control (CDC) estimates that 20.4% of adults have chronic pain [3], and 4.8% to 7.4% of adults have “high-impact chronic pain” defined as severe pain that causes disability and limits life or work activities [3, 5]. Chronic pain is highest among adults over 65 (30.8%), women (21.7%) and non-Hispanic white adults (23.6%) [3]. In the United Kingdom, the prevalence of chronic pain ranges from 35% to 51.3% of adults with 10.4% to 14.3% having moderate or severely disabling chronic pain [6]. Internationally, the prevalence and demographics of chronic pain is consistent with epidemiological studies from the US and the UK [7–16]. However, other countries had mixed results and did not observe a higher prevalence in older age groups, did not demonstrate sex differences or had a higher prevalence in males [17–19].

Pain is not only influenced by biological factors, but also psychosocial factors due to the overlap of physical, emotional, or behavioral supraspinal pain pathways [20]. As a result, pain is also highly associated with psychiatric comorbidities resulting in more severe pain, worse functioning and disability, lower quality-of-life, and higher incidences of opioid-use disorder [21]. The most prevalent psychological comorbidity is depression or major depressive disorder (MDD) (30–60%) [22–24]. Other disorders include bipolar disorder, panic disorder, post-traumatic stress disorder, anxiety, obsessive-compulsive disorder, sleep disorders and specific phobias [24]. Antidepressants such as serotonin norepinephrine reuptake inhibitors (SNRIs) and tricyclic antidepressants are often prescribed to those in chronic pain, which can also help those with comorbid depression. Other medication therapies that can treat both conditions include N-methyl-D-aspartate receptor (NMDAR) antagonists (e.g., ketamine) and cannabinoids. In addition, non-pharmacotherapies include behavioral interventions such as cognitive behavioral therapy (CBT). Originally developed to treat depression, CBT is now also utilized for chronic pain and focuses on the relationship between cognition or thoughts, emotions, and behaviors and how they relate to pain functioning [25].

While there are several types of chronic pain disorders, the most resistant to current pharmacologic therapies is neuropathic pain (NP). The prevalence of NP is poorly understood but is estimated to be 6.9% to 10% of the population, with the highest occurrence in black and Hispanic adults [26–29]. In addition, the financial burden of chronic pain in the United States is estimated to be as high as $635 billion a year, which is more than the annual cost for cancer and heart disease combined.[30]

NP is a collection of pain disorders resulting from damage to the somatosensory nervous system [31], characterized by disease or lesions on the neurons that can occur peripherally (e.g., chemotherapy-, diabetic- or postherpetic-induced neuralgias) or centrally (e.g., spinal cord injury (SCI) or post-stroke pain) [32]. Unlike nociceptive or inflammatory pain, the symptoms of NP include electrical shocks, allodynia, thermal hypersensitivities and burning paresthesias; these symptoms are spontaneous, paradoxical, and refractory to traditional analgesic therapies [33]. Though the symptoms between various NP conditions are similar, generalizing therapies to every NP condition remains a challenge as the mechanisms, underlying neuropathology, and etiologies may be vastly different.

NP is pharmacologically treated with non-opioid and/or opioid medications [34]. Non-opioid therapies such as antidepressants and anticonvulsants are often prescribed but are not completely effective [35, 36]. Those who resort to opioid therapy for relief are exposed to many dangerous side effects limiting their use, particularly respiratory depression, addiction, accidental overdose, and death [37]. Multimodal therapy (i.e., a regimen of drugs from different drug classes) is often recommended as single-drug therapy may not provide satisfactory analgesia [38, 39]. Additionally, there are numerous challenges associated with clinical studies of NP, predominantly surrounding the placebo response rates, which have been increasing steadily over recent years [40]. Multiple clinical trials have failed to show a beneficial effect for drugs that have promising analgesic activity. Data from randomized controlled trials of drugs for the treatment of NP between 1990 and 2013 have shown that the placebo response rates have increased while responses to the drug have remained stable over this period of time resulting in reduced separation between drug effect and placebo [40] and the failure to show efficacy. There are multiple reasons suggested for this outcome, along with solutions to address the difficulty. These include concerns related to the lack of reliability surrounding subjective assessment of efficacy. Functional neuroimaging has been used to identify neural changes associated with placebo response that differ from the response to the drug and which, if extended, may provide useful information [41]. Analysis of the placebo responses in clinical trials of depression also point to several implications for NP that include the exclusion of patients with mild pain severity, exclusion of patients with short episode duration along with minimizing the number of treatment groups and extraneous contact with the investigative staff and other sources of nonspecific therapeutic effects [42]. Additionally, Finnerup et al. [43] and Arakawa et al. [44] have identified a number of potential contributors to the elevated placebo response that include the study duration and size, as well as the study design and number of patients at a particular site that should be considered when planning and conducting confirmatory trials in NP. All these issues point to the fact that the failure to translate findings from preclinical models of pain may not always be the shortcomings of the animal models, and that the failure rate of drugs for NP requires a full and detailed evaluation of both preclinical and clinical approaches to this problem.

Despite the tremendous efforts to understand mechanisms associated with the development of NP, the identification of new pain targets and the development of novel pain therapeutics remains limited. The discovery of effective, non-addicting analgesic therapies to treat NP is essential to combat this devastating disease.

2. Current Non-opioid Therapies for Neuropathic Pain

Although opioids are prescribed for pain, concerns regarding poor efficacy, tolerance, dependence, and overdose limit their utility. Some patients remain heavily reliant on opioid therapy to adequately treat their pain but the strong addictive properties of opioid-based analgesics place patients at risk for developing opioid use-disorder. Opioid-use disorder has steadily increased with the rise of synthetic opioids leading to a dramatic surge in overdose deaths. Deaths from this opioid crisis has reached an all-time high within the United States, in part due to the Covid-19 pandemic. Pre-pandemic in 2019, the CDC reported 71,000 drug overdose deaths, over 70% of which involved opioids such as fentanyl, oxycodone, hydrocodone and heroin [45]. One year later, drug overdose deaths had risen to 92,000 people - a shocking 29% increase from the year prior. In 2021, opioid-related deaths reached nearly 100,000 people, the highest number in 30 years [46].

Treatment for NP includes many non-opioid alternatives; however, the majority of these drugs are not completely effective and also have a range of side effects. For example, gabapentin, a common anticonvulsant used for NP, is only effective in 30–35% of users [47, 48]. Based on available evidence, NP treatment guidelines are created by international pain associations, including the European Federation of Neurological Societies (EFNS) [49], IASP’s Neuropathic Pain Special Interest Group (NPSIG) [50], the National Institute for Health and Care Excellence (NICE) [51], and the Canadian Pain Society [52]. The recommended first-line therapies for NP include tricyclic antidepressants (TCAs), serotonin-norepinephrine reuptake inhibitors (SNRIs), and gabapentinoids such as pregabalin and gabapentin. In trigeminal neuralgia, though, the first-line therapy is carbamazepine [53]. Recommended second-line therapies are lidocaine patches, capsaicin patches, or tramadol. Lastly, third-line treatments include opioids, botulin toxin A and other anticonvulsants [38, 54–56]. Of note, while acetaminophen and oral NSAIDS (i.e., ibuprofen) are commonly used for nociceptive and inflammatory pain, they are not effective for the treatment of NP [51, 57, 58].

NP patients often are prescribed two or more medications as combination therapy to increase efficacy. For example, gabapentinoids and antidepressants had better efficacy in multimodal therapy compared to monotherapy.[59] In addition, a clinical trial to evaluate gabapentin plus donepezil (a reversible acetylcholinesterase inhibitor used to treat dementia from Alzheimer’s disease) in an experimental model of electrical hyperalgesia suggested further clinical investigation of a gabapentin-donepezil combination in patients with an inadequate response to gabapentin alone [60]. Some consider multimodal therapy may also allow lower doses of each medication, decreasing the incidence of side effects. However, while two-drug combinations provided better efficacy, many trials reported significant drop-outs due to increased side effects [61].

2.1. Antidepressants

As first-line therapeutics, antidepressants such as serotonin norepinephrine reuptake inhibitors (SNRIs) and tricyclic antidepressants (TCAs) are often prescribed to those with NP. In the descending or modulation pathway, pain is inhibited by serotonergic or noradrenergic neurons. These pathways start in the brain or brain stem and ultimately control the release of inhibitory GABA and endogenous opioids in the spinal cord to decrease pain transmission. Thus, antidepressants work to block the reuptake of serotonin and norepinephrine from the synapse to increase the activity of pain attenuation. In addition to treating somatic physical pain, antidepressants also benefit patients by treating psychological comorbidities.

2.1.1. SNRIs

There are several SNRIs that have been studied in NP including duloxetine [62, 63], venlafaxine [64] and milnacipran [65]. In a systematic review and meta-analysis (33 studies, n=5318), there was moderate evidence that SNRIs reduced back pain at 3–13 weeks, though it was not clear if these patients had neuropathic symptoms. In addition, there was evidence that SNRIs were efficacious for sciatica and osteoarthritis [65]. Duloxetine has demonstrated efficacy in diabetic peripheral neuropathy and fibromyalgia but has not been well studied in other types of NP. Side effects are mild (e.g., nausea) and duloxetine lacks major cardiovascular adverse effects sometimes seen with TCAs. In a systematic review (18 studies, n=6407), there was sufficient evidence to support that duloxetine (60–120 mg) provides greater than 50% pain relief at 12 weeks in those with diabetic neuropathy [63]. Participants with fibromyalgia had similar pain relief at 12 and 28 weeks, however, due to small sample sizes and exclusively manufacturer-led studies, more unbiased, investigator-led trials are needed to support the use of duloxetine for this condition [63, 66]. Venlafaxine has also been shown to be effective in polyneuropathies but should be used with caution in those with cardiac comorbidities as cardiac conduction abnormalities and hypertension was identified in some patients [67]. In a review of 6 randomized controlled trials (n=460), venlafaxine (150–225 mg) was primarily studied in diabetic peripheral neuropathy and was found to have some positive benefit in 4 out of 6 trials. Somnolence, dizziness, and mild gastrointestinal problems were the major adverse effects; however, this was not significantly different from the placebo groups. These studies had strong placebo effects and venlafaxine was not recommended over other treatments.[68] In chemotherapy-induced peripheral neuropathy (CIPN), duloxetine had superior efficacy compared to venlafaxine [69]. Milnacipran has only been studied in fibromyalgia with minimal or no difference in pain scores compared to placebo [70–72].

Tramadol, a weak opioid agonist, also has SNRI activity and has demonstrated some efficacy in NP [73, 74]. Because tramadol also inhibits the reuptake of serotonin and norepinephrine, it can have severe adverse drug interactions with antidepressants causing a rare but potentially fatal reaction called serotonin syndrome [75, 76]. Some believe the risk of addiction with tramadol is considerably less than other opioids such as morphine and oxycodone [77, 78] but other reports have not supported this claim [79, 80] and, thus, tramadol continues to be a schedule IV drug in the US [81]. In addition, because of its opioid properties, withdrawal after discontinuation is a major adverse effect [82].

2.1.2. TCAs

Examples of TCAs include amitriptyline[83], nortriptyline[84], and desipramine[85] which are often prescribed but have many off-target effects causing cardiac toxicity and anticholinergic side effects (i.e., urinary retention, dry mouth, constipation and orthostatic hypotension) [86]. In a systematic review and meta-analysis (17 studies, n=1342) including seven NP conditions, only third-tier evidence (obtained from well-designed controlled trials without randomization) was available, and only 2 trials determined amitriptyline was more efficacious than placebo. In addition, more patients who took amitriptyline experienced adverse events than the placebo groups. The authors concluded amitriptyline will not work for most people with NP, although no studies met the current standards of evidence [83].

2.1.3. SSRIs

Selective serotonin reuptake inhibitors (SSRIs) have replaced TCAs as first-line treatment of depression due to a better safety profile, however, SSRIs are less efficacious in NP compared to TCAs or SNRIs and are considered third-line therapy. Animal models of pain also demonstrate better antinociceptive effects from antidepressants that affect both serotonin and norepinephrine levels [87]. In humans with diabetic peripheral neuropathies, citalopram and paroxetine has shown some benefit, whereas fluoxetine has no efficacy [88]. Escitalopram has also shown to provide pain relief in patients, but the evidence is not strong enough to be recommended for first- or second-line therapy in NP [89].

2.2. Anticonvulsants

2.2.1. Gabapentinoids

Gabapentinoids are considered first-line therapy for NP but with limited efficacy [33, 90]. Gabapentin and pregabalin are anticonvulsants that bind to the α2-δ subunit of voltage-gated calcium channels to reduce calcium influx, inhibit neurotransmitter release and prevent pain transmission [91]. Both drugs structurally resemble GABA but have no activity in GABAergic systems. A meta-analysis suggested that gabapentinoids are effective in decreasing NP and other secondary outcomes after SCI [92]. A large meta-analysis of 37 studies (n=5633) was performed studying oral gabapentin (1200 mg or greater) in 12 chronic pain conditions mostly consisting of postherpetic neuralgia, painful diabetic neuropathy, or mixed NP. Only 35% of participants achieved at least 50% pain intensity reduction compared with 21% for placebo. Furthermore, over half of participants did not achieve “worthwhile pain relief” [47, 48]. In those with diabetic peripheral neuropathy, gabapentin and duloxetine were equally effective, however, gabapentin showed an earlier effect but more side effects while duloxetine had better medication compliance [93]. In addition, there are several adverse effects associated with gabapentin including sedation, dizziness, peripheral edema, and gait disturbances, and it must be used with caution in those with renal insufficiency.

Another gabapentinoid used to treat NP is pregabalin, which has a slightly different pharmacokinetic profile from gabapentin that include increased absorption, quicker peak times and greater bioavailability. Pregabalin also has an increased binding affinity to calcium channels and higher potency[94]. In preclinical studies in animal models of NP, pregabalin decreases symptoms of allodynia and hyperalgesia. Pregabalin has also been used to explore various molecular targets and signaling systems including the activation of excitatory amino acid transporters (EAATs), potassium channels and inhibition of pathways involving inflammatory mediators [95]. Clinical studies in different age groups and in different types of NP (i.e., peripheral diabetic neuropathy, fibromyalgia, post-herpetic neuralgia, CIPN) have projected pregabalin as the most effective agent either as monotherapy or in combined regimens in terms of cost effectiveness, tolerability, and overall improvement in NP states. Pregabalin has been shown to provide pain relief more quickly than gabapentin. In NP patients who are refractory or intolerant to gabapentin, pregabalin may be used successfully [96]. As of 2019, pregabalin and gabapentin were reclassified as Schedule III drugs in the United States, which are drugs with moderate to low potential for psychological and physical dependence. At high doses, gabapentin may induce euphoria and withdrawal after several weeks of treatment leading to widespread recreational misuse and drug diversion.[97]

2.2.2. Other anticonvulsants

Other anticonvulsants used for pain are valproate [98–102] (or valproic acid; VPA) and riluzole, a drug commonly used for amyotrophic lateral sclerosis (ALS) [103–105]. Valproate and riluzole also prevent the release of glutamate through the blockage of inactivated sodium channels. Unfortunately, clinical trials with VPA for NP have demonstrated low efficacy and high adverse effects [98], and high doses of riluzole were not effective in alleviating peripheral NP [106]. Because of these adverse events, anticonvulsants other than gabapentin and pregabalin are third- or fourth-line therapy.

2.3. NMDA Receptor Antagonists

During pain transmission, glutamate is released from afferent terminals and stimulates postsynaptic glutamate receptors to mediate excitatory neurotransmission. In nociceptive transmission, excessive activation of glutamate receptors, such as NMDARs, occurs through three mechanisms: (1) increased glutamate release from terminals in dorsal horn, (2) increased number and function of glutamate receptors, and (3) decreased glutamate clearance [90]. To decrease pain transmission, NMDAR antagonists are used intermittently for the treatment of NP. Currently, there are 8 NMDAR antagonists available: ketamine, carbamazepine, methadone, memantine, amantadine, dextromethorphan, phenytoin and VPA [90, 107, 108]. Because some of these drugs have multiple mechanisms of action, they are also classified as opioids, antidyskinetics, anticonvulsants or cough suppressants.

2.3.1. Ketamine

In a systematic review of NMDAR antagonists for the treatment of NP, intravenous (IV) ketamine was the most efficacious and provided adequate analgesia in 13 clinical trials [90]. This result was also consistent amongst various NP pathologies including SCI, phantom limb pain, and complex regional pain syndrome (CRPS). Several studies, including a Phase II open label study in refractory CRPS patients [109] and a Phase II double blind placebo-controlled study in outpatients with CRPS [110] have been conducted with IV ketamine. Both studies reported benefits in pain reduction, as well as in the quality of life and ability to work. A meta-analysis of the effect of ketamine in the treatment of CRPS suggested that ketamine infusion can provide clinically effective pain relief but that the high heterogeneity of the studies warrant a random controlled trial that includes multicenter studies [111], a conclusion also from the studies conducted by Schwartzman and colleagues [109, 110]. IV ketamine administration may require several days of infusion under the supervision of an anesthesia-trained clinician for safety. In addition, the use of ketamine for NP is not FDA approved and currently considered an experimental therapy.

Ketamine strongly inhibits neuronal hyperexcitability by binding noncompetitively to the phencyclidine site of NMDARs through allosteric modulation. Though supplied as a racemic mixture, the S (+) enantiomer of ketamine has higher clearance, greater therapeutic index and potency, and a more favorable side effect profile (i.e. less lethargy, impairment, psychotic reactions and agitation) [112]. Ketamine undergoes liver metabolism to an active metabolite (nor-ketamine) that also has analgesic properties.

2.3.2. Carbamazepine

Other NMDAR antagonists used for the treatment of pain are carbamazepine and methadone. Carbamazepine, an antiepileptic medication, blocks NMDAR currents and inhibits Ca²⁺ influx preventing NMDA-mediated toxicity. Carbamazepine showed clinical efficacy for NP in a variety of conditions including trigeminal neuralgia, complex regional pain syndrome (CRPS), diabetic neuropathy and SCI [90]. However, this medication has significant side effects including sedation, poor concentration, and an impact on psychomotor functions [113, 114]. More serious adverse effects also include hepatotoxicity, Steven-Johnson syndrome, and toxic epidermal necrolysis.

2.3.3. Methadone

Methadone is a long-acting opioid agonist that also has NMDAR antagonist activity. Clinically available as a racemic mixture, the (L) isomer of methadone has the opioid activity, while the (D) isomer has the NMDAR activity. Unlike ketamine, however, methadone does not produce active or toxic metabolites. There are trials that support the use of methadone for the treatment of NP [115, 116], however, due to its long half-life and slow onset, it takes approximately 3 to 5 days to achieve analgesic effects. The major side effects associated with methadone is QT prolongation leading to cardiac complications.

2.3.4. Other NMDAR antagonists

Unfortunately, other NMDAR antagonists such as memantine, amantadine, dextromethorphan, phenytoin and VPA had inconsistent results or severe adverse effects and are not recommended for the treatment of NP [90]. VPA is also reported to modulate levels of EAATs, and therefore is further discussed below in the section on drugs that increase EAAT expression.

2.4. Topical & Localized Therapies

Topical agents offer advantages as a safe alternative to systemically administered drugs in those with chronic, localized NP. Local delivery of medications via patch or creams provides more site-specific effects, less first-pass metabolism, lower drug-drug interactions, and less systemic adverse effects, especially in the elderly [117, 118].

2.4.1. Lidocaine

Local anesthetics such as lidocaine are sodium channel blockers that slow the propagation of sensory action potentials in pain transmission. Though local anesthetics are commonly administered via local, neuraxial or systemic injection, it is the topical formulations (5% lidocaine patch and gel) that offers the most benefit in certain NP conditions. Well-localized NP pain conditions such as postherpetic neuralgia, HIV-induced neuropathy and diabetic peripheral neuropathy appear to benefit the most from topical lidocaine therapy compared to those with central NP disease[119]. Due to minimal systemic absorption, lidocaine patches have few adverse effects or drug interactions. In a systematic review and meta-analysis (12 studies, n=508) lidocaine in various formulations (5% patch, 5% cream, 5% gel, 8% spray) was compared to active controls and placebo. In this review, most participants experienced postherpetic neuralgia followed by trigeminal neuralgia, post-surgical neuralgia, or post-traumatic neuralgia. There was no first- or second-tier data in this systematic review, however, third-tier showed lidocaine had some benefit over placebo with no major adverse events [119].

2.4.2. Capsaicin

Capsaicin is a chemical compound in chili peppers that stimulates TRPV1 receptors, a thermal nociceptor commonly found in the cutaneous tissue. Topical application of capsaicin initially causes enhanced sensitivity, followed by reduced sensitivity, then persistent desensitization with repeated applications. Low-concentration oral dosing may be used, but the high-concentration patch (8%) appears to have better efficacy that can last up to 12 weeks. However, the long-term effects of this therapy are unknown and adverse reactions may include mild application-site reactions such as erythema, or severe effects such as permanent epidermal denervation [120]. In an analysis of 8 studies (n=2488), participants with localized NP conditions (i.e., post-herpetic neuralgia, HIV-induced neuropathy, diabetic peripheral neuropathy) experience more pain relief from their baseline compared to control (0.04% capsaicin) that persisted for 2–12 weeks. However, this therapy was not effective for persistent post-surgical pain. Local side effects were common but not consistently reported, and the adverse events did not differ significantly between groups. Overall capsaicin provided moderate or substantial pain relief compared to control [120]. Another large systematic review assessed 108 studies and found similar results that capsaicin improves localized NP without affecting motor and large nerve fibers involved in other sensory functions [121].

2.4.3. Botulinum toxin

Known for causing botulism in humans, botulinum toxin is being investigated as a possible pain therapy, although its only approved pain-related use is for migraines [122]. Botulinum toxin can cause flaccid paralysis of skeletal muscles and dysautonomia via inhibition of acetylcholine release at neuromuscular junctions. Antinociceptive mechanisms are not as well understood but the toxin may prevent glutamate release, and interfere with Na⁺ channels, mechanosensitive receptors, and transient receptor potential ankyrin 1 (TRPA1) channels [123]. Though botulinum toxin has been successfully utilized in many neuromuscular conditions, there have been inconsistent results in randomized controlled trials (RCTs) of chronic pain. Smaller RCTs showed significant pain reduction in a 12-week period however, larger trials showed no differences compared to placebo [33]. Currently, there is no strong evidence to recommended botulinum toxin for the routine treatment of NP.

2.4.4. Other topical agents

Other topical agents have been studied in smaller trials or reported as case studies with no controls. These agents include topical ketamine, clonidine, gabapentin, baclofen, phenytoin, isosorbide dinitrate, amitriptyline and loperamide. In a small 4-week RCT (n=29), cannabidiol oil (CBD) was studied in peripheral neuropathy of the lower extremities with a reduction in intense pain, cold and itching sensations, without adverse events [124]. For topical agents in central NP syndromes, such as SCI, only case studies have been reported [125].

2.5. Cannabis & Cannabis-based Medicines (CBM)

Medicinal cannabis and synthetic cannabinoids are increasingly used to manage chronic pain, however, there is limited understanding of their efficacy and safety in NP disorders. The endocannabinoid system (eCS) consists of cannabinoid receptors 1 and 2 (CB₁R, CB₂R), endogenous cannabinoid ligands (endocannabinoids) and its metabolizing enzymes. The eCS has broad roles in stress recovery, immune regulation, cognition and memory, nausea and vomiting, and antinociception [126]. CB₁R and CB₂R are located at peripheral, spinal and supraspinal sites, and agonists to these receptors (endocannabinoids, phytocannabinoids or synthetic cannabinoids) have antinociceptive effects in animal models of acute, inflammatory and NP [127]. Proposed mechanisms for the role of the eCS in antinociception include modulation of postsynaptic neuronal excitability, inhibition of presynaptic neurotransmitter release, activation of descending inhibitory pathways and reductions in neuroinflammatory signaling. Despite promising results in animal studies, the effects of cannabis-based medicines (CBM) in clinical trials have not shown robust differences between treatment and placebo in acute pain [128, 129] or NP conditions [130].

Though some NP patients report that CBMs are effective in reducing pain, many RCTs do not support this claim. A review of 16 randomized, double-blind control trials (1750 participants) studied the effects of plant-derived or synthetic CBM against placebo for efficacy and safety [130]. CBM was administered via oromucosal spray of tetrahydrocannabinol (THC) and/or cannabinol (CBD); nabilone, a synthetic cannabinoid; inhaled herbal cannabis; and dronabinol, a plant-derived THC. A variety of NP conditions were studied including cancer-related neuropathy, central NP (multiple sclerosis, SCI), CRPS, HIV neuropathy, diabetic peripheral neuropathy, phantom limb pain and trigeminal neuralgia (excluding fibromyalgia). Efficacy was measured as pain relief of 30–50% or greater, and a global impression of pain to be “much improved” or “very much improved”. Only 21% of participants reported their pain relief was 50% or greater after CBM compared to 17% of those who received placebo, and there was no significant improvement in global pain impression (28% of CBM vs. 22% of placebo). In addition, more study participants from the CBM groups withdrew from studies due to adverse events compared to placebo. There was moderate to low quality evidence for pain-relief outcomes because of small sample sizes and exclusion of participants with a history of substance-use disorder or other medical comorbidities. In conclusion, this systematic review found no significant benefits of CBM for chronic NP and any potential benefits may be outweighed by potential harm [130].

Besides questionable efficacy, chronic users of cannabis and CBM may be exposed to several adverse effects. In addition to cognitive alterations, airway disease and heart disease [131], cannabinoids may also negatively interact with the glutamatergic system to produce psychotic disorders including mood disorders and schizophrenia [132]. These pro-psychotic effects are associated with central CB₁R agonists, which modulate the levels and functions of neurotransmitters like dopamine, serotonin and glutamate in the hippocampus, frontal cortex, and cerebellum [133, 134]. Due to these adverse effects, there is an increased focus on strategies to dissociate the psychoactive effects of cannabinoids from its potential analgesic effects. Examples include restricting CB₁R agonists to peripheral receptors, creating specific CB₂R agonists, inhibition of endocannabinoid metabolism or reuptake and modulating non-CB₁R/CB₂R targets of cannabinoids such as TRPV1, PPARs and GPR55 [127]. Altogether, high-quality trials with large sample sizes are required to demonstrate any efficacy of CBMs for NP while minimizing its negative effects.

3. The Glutamatergic System and Neuropathic Pain

In the mammalian nervous system, nociception occurs through a complex interaction of receptors, transporters and neurotransmitters on neurons and glia that use L-glutamate, the major transmitter released by sensory afferents in the nervous system. Glutamate is critically involved in nociceptive signaling. Elevated levels of extracellular glutamate contribute to worsened symptoms and NP progression [135, 136]. In addition, the expression of glutamate receptors and transporters becomes dysregulated in chronic pain conditions. Understanding the regulation of the glutamatergic system both physiologically and after nerve injury will guide the development of pain-relieving pharmacotherapies [135, 137].

3.1. Neurotransmitters

There are many neurotransmitters and peptides involved in nociception and pain modulation including substance P, calcitonin gene-related peptide, gamma aminobutyric acid (GABA), glycine and the main excitatory neurotransmitter, glutamate [138, 139]. Of the neurotransmitters involved in pain, glutamate plays a crucial role and is present in peripheral and central afferent fibers. In addition to nociception, glutamate signaling also stimulates a variety of complex tasks involved in neuronal development, control of movement, synaptic plasticity, learning and memory and neuronal cell death [140, 141]. Glutamate has also been found in motor neurons, particularly in the ventral horn of the spinal cord, and can be found in the soma, axons and terminals of both sensory and motor neurons [142]. While glutamate is essential for many biological processes, excess release in the central nervous system (CNS) or peripheral nervous system (PNS) can lead to pathological conditions including excitotoxic cell damage in neurons and glia.

During normal excitatory signaling, glutamate undergoes a glutamate-glutamine cycle (Figure 3) between glia and neurons. Upon neuronal stimulation, glutamate is released into the synaptic cleft and stimulates fast-acting ionotropic receptors and slower-acting metabotropic receptors [90]. To terminate the activation of glutamate receptors, extracellular glutamate must diffuse away from the synaptic cleft or undergo reuptake through glutamate transporters (EAATs) located in the cell membranes of neurons and glia. Though microglia and oligodendrocytes in the CNS contain glutamate transporters, astrocytes play the largest role in removing extracellular glutamate. After astrocytic reuptake, glutamate is converted into glutamine via glutamine synthetase, an enzyme highly expressed in astrocytes, to maintain continuous glutamate uptake from the extracellular space [143]. Then, glutamine is transported out of the glial cells by the system N transporter 1 (SN1) and back to the presynaptic neuron terminal through system A transporter 2 (SAT2). In the presynaptic neuron, glutamine is then metabolized by a mitochondrial enzyme, glutaminase [144], back into glutamate. Once in the presynaptic terminal, glutamate is then packaged into vesicles through vesicular glutamate transporters (vGLUTs 1–3)[145]. Upon stimulation, such as an action potential, these vesicles fuse with the cell membrane and release glutamate into the synapse to repeat this cycle.

Figure 3: The glutamate-glutamine cycle in the tripartite synapse.

In glutamatergic signaling, the canonical tripartate synapse is composed of the presynaptic neuron, postsynaptic neuron and the astrocytic endfoot. The cell membrane of each terminal contains various receptors that participate in excitatory potentials, and transporters that shuttle glutamate to maintain homeostasis. An action potential in the presynaptic neuron stimulates the release of glutamate into the synaptic space (1) and stimulates pre and postsynaptic glutamate receptors (mGluRs and iGluRs). Glutamate diffuses away from the synapse and undergoes uptake into the astrocyte through EAATs (2). EAAT2 is responsible for the majority of glutamate uptake in the CNS. In excitotoxicity conditions, glutamate may diffuse into the extrasynaptic space and stimulate extrasynaptic glutamate receptors. Glutamate in the astrocyte is rapidly converted to glutamine-by-glutamine synthetase (3). Glutamine is exported through system N transporter 1 (SN1) to the extracellular space then through system A transporter 2 (SAT2) into the presynaptic neuron (4). In the presynaptic neuron, glutamine is converted back to glutamate by glutaminase (5). Glutamate is transported into vesicles through vesicular glutamate transporters 1 or 2 (vGLUT1–2, 6). An action potential stimulates the fusion of the vesicle with the cell membrane for neurotransmitter release.

Not shown: presynaptic EAAT2, EAAT5; cystine/glutamate antiporter (xCT).

Created with BioRender.com

In addition, glutamate reuptake into GABAergic neurons serves as a precursor for the synthesis of the inhibitory neurotransmitter GABA, which helps to attenuate pain [142]. Other neurotransmitters involved in the descending modulation pathway include serotonin, norepinephrine, and endogenous opioids (i.e., enkephalins, endorphins, dynorphins).

3.2. Glutamate Receptors

Glutamate receptors (GluRs) are a heterogenous protein population that integrate a variety of signals such as ions, endogenous ligands, membrane voltage and intracellular proteins to regulate receptor stimulation, location, and expression [146]. While GluRs are traditionally thought to be located in pre-synaptic and post-synaptic neurons, they are also found on peripheral glial cells (e.g. Schwann cells) to communicate with neuronal axons or other glia [142]. GluRs in neurons and glia are divided into two major classes: ionotropic (iGluRs) and metabotropic (mGluRs).

3.2.1. Ionotropic Glutamate Receptors (iGluRs)

iGluRs facilitate responses at most excitatory synapses and are known to mediate fast postsynaptic potentials though ligand-gated ion channels, though a metabotropic function of certain iGluRs has recently been elucidated [147–149]. iGluRs can be further broken down into four families named by the agonist originally used for their discovery. They include AMPARs, NMDARs, kainate receptors, and delta receptors. All iGluRs are tetramers of different subunits to form an ion channel and each receptor can be categorized by different subunits leading to further complexity and heterogeneity.

NMDARs allow the passage of Na⁺, K⁺, and Ca²⁺, which generates intracellular signaling processes in normal conditions and excitotoxicity in pathological states such as NP. The role of iGluRs in pain is well reviewed in [150] and [151]. Of note, a study in the rat spinal nerve ligation (SNL) model observed that injection of MK-801, a non-competitive NMDAR antagonist, delayed the onset of paw withdrawal to stimuli, whereas other receptor-subtype antagonists had no effect, suggesting that the manipulation of NMDARs reduces NP [152]. When extracellular glutamate levels are elevated with glutamate transporter blockers (TBOA and dihydrokainate), spontaneous nociceptive behavior is observed in rodents, such as shaking, licking, or biting. These behaviors were significantly attenuated by the administration of NMDAR antagonists (MK-801 and AP-V). These results suggest NMDAR activation plays a role in nociceptive behaviors [153]. Some NMDAR antagonists (such as ketamine, carbamazepine, and others) are currently used in the clinic and are discussed above (section II).

3.2.2. Metabotropic Glutamate Receptors (mGluRs)

mGluRs are G-protein coupled receptors (GPCRs) that facilitate slower synaptic potentials and signal through complex intracellular pathways or other ion channels[154]. They are divided into eight functionally distinct subtypes (mGluR1–8) that exist as constitutive dimers. These subtypes are classified into three groups (Group I-III) according to their signal transduction pathways, sequence identity and pharmacology[154]. Group I (mGluR1 & mGluR5) are Gαq GPCRs coupled to phospholipase C (PLC) that ultimately leads to intracellular calcium signaling. Group II (mGluR2 & mGluR3) and Group III (mGlur4, 6, 7 & 8) are Gαi GPCRs that inhibit adenylate cyclase and decrease cAMP [146, 154]. Physiologically, they generate slower post-synaptic potentials than iGluRs and are implicated in long-term changes, particularly in synaptic plasticity and the formation of learning and memory [154]. All mGluR subtypes (except for mGluR6, which is localized in the retina only) are expressed within the nociceptive pathways where they play roles modulating pain transmission[135, 155]. Many selective compounds for the different mGluR subtypes have been identified in the last few years, contributing to our understanding of their specific roles in nociception and suggesting that these receptors are promising targets for the treatment of chronic pain. However, concerns about motor and cognitive impairment associated with mGluR1 antagonism [156–158], psychoactive properties with mGluR5 antagonism [159] and tolerance with group II agonism [160, 161], suggest caution in the development of therapies based on mGluR modulation. Nonetheless, targeting peripheral mGluRs has shown analgesic effectiveness, which suggest that spatiotemporal tuning of these receptors as a strategy with reduced off-target effects associated with systemic modulation of the glutamatergic system [162].

3.3. Glutamate Transporters

Glutamate signaling is terminated by the diffusion of glutamate away from the synapse, or removal of extracellular glutamate by five distinct subtypes of EAATs, which belong to the solute carrier family 1 (SLC1) group of membrane transport proteins. EAATs are approximately 500–600 amino acid residues long, have 50–60% homology, and are comprised of six to eight transmembrane domains, one to two re-entrant loops, and a cytoplasmic N- and C-terminus. The EAATs are secondary-active transporters that couple the movement of one glutamate with the symport of three Na⁺ ions and one H⁺ and the counter transport of one K⁺ [163]. The Na⁺ gradient generated by Na⁺ K⁺- ATPase activity drives glutamate transport under physiological conditions [164].

EAATs are essential for the fast removal of glutamate from the extracellular space upon its release into the synaptic cleft, keeping the extracellular glutamate concentrations low (25–600 nM) [165, 166]. Elevated concentrations of glutamate to approximately 2–5 μM, which can occur during ischemia or injury such as of traumatic brain injury, results in excitotoxic damage [167–170]. Glutamate excitotoxicity is the process in which failure to remove extracellular glutamate from the synaptic cleft results in sustained extracellular glutamate levels and excessive activation of NMDARs [169], resulting in high Ca²⁺ influx into the postsynaptic neuron leading to downstream activation of a cascade of phospholipases, endonucleases, and proteases like calpain that leads to cell apoptosis [171]. Additionally, the excessive Ca²⁺ influx and subsequent depolarization of the postsynaptic neuron may lead to the efflux of excessive glutamate further worsening excitotoxicity [172]. The imbalance of release and uptake of glutamate by alterations in the function and expression of these transporters may cause disturbances in neuronal signaling, leading to neurological and psychiatric disorders, such as stroke[173, 174], epilepsy[175, 176], traumatic brain injury[177–180], Huntington’s disease [181], Alzheimer’s disease[182, 183], ALS[104, 184], and NP [185, 186].

EAATs are classified into (rat/human homologue): GLAST/EAAT1, GLT-1/EAAT2, EAAC1/EAAT3, EAAT4 and EAAT5. Throughout this review, we will use this nomenclature interchangeably, according to the original cited article. The majority of EAATs expressed on astrocytes are EAAT1 and EAAT2, which have similar structures and affinity for glutamate but differ in their expression profile in the CNS. In early development, EAAT1 is highly expressed in progenitor cells such as radial glial cells [187]. In the mature brain, EAAT1 becomes highly expressed in the cerebellum (Bergmann glial cells), retina (Müller cells), cochlear glial cells in the inner ear and circumventricular organs, whereas their expression in the forebrain is limited [143, 188, 189]. However, after CNS maturation, EAAT2 represents the majority of EAATs in astrocytes and oligodendrocytes of the brain and spinal cord[190], removing most of the extracellular glutamate[191, 192]. EAAT2 is also expressed in neurons [193, 194] at a lower extent than in astrocytes in the presynaptic axon terminals. Overall, EAAT2 accounts for approximately 95% of the total glutamate transport activity and 1% of the total brain protein in the CNS [192, 195–198] and, thus, plays a central role for maintaining the homeostasis of glutamate [199–202]. When EAAT2 function is impaired or altered near afferent synapses, neuronal and glial pathological changes occur leading to painful hypersensitivities [203, 204]. Therefore, EAAT2 is of particular interest in the study of NP development [185].

EAAT3 is mostly expressed in postsynaptic neurons, where it buffers nearby glutamate receptors and modulates excitatory neurotransmission and synaptic plasticity in the brain [205]. The neuronal EAAT4 and glial EAAT1 are the predominant EAATs that prevent neurotoxicity by maintaining low extracellular glutamate levels in the cerebellum to limit mGluR1 signaling [206]. EAAT5 mediates glutamate release and light responses in depolarizing bipolar cells in the retina[207]. In addition, EAATs are also present on GABAergic terminals to remove glutamate, which is converted into GABA by glutamic decarboxylase (GAD). This function limits the excess activation of extra-synaptic and neighboring synaptic glutamate receptors [208].

In addition to the EAAT family of transporters and vesicular glutamate transporters, there are also glutamate-cystine exchangers (SLC7A11, xCT) present in neurons and glia that exchange intracellular glutamate for extracellular cystine in a sodium-independent way, to provide a cystine source for the synthesis of glutathione [209]. Extracellular accumulation of glutamate can also cause toxicity through interaction with these exchangers as they become blocked and glutathione cannot be synthesized, leading to oxidative stress and downstream cell death [210]. xCT was shown to modulate cancer-induced bone pain [211, 212], however, further studies are needed to determine whether the modulation of this transporter can yield translational strategies for NP treatment.

Glutamate transporter inhibitors are frequently used to analyze the role of EAATs in various cellular mechanisms or disease processes, including pain [213]. These compounds include DL-threo-β-benzyloxyaspartate (DL-TBOA), pyrrolidine-2,4-dicarboxylic acid (PDC) and dihydrokainate (DHK) [214, 215]. In normal, uninjured animals, EAAT inhibitors increase extracellular glutamate concentrations by blocking the reuptake of synaptic glutamate and, thus, produce hyperalgesia. Conversely, animals exposed to IL-1β, formalin or Complete Freund’s Adjuvant (CFA) experience anti-nociception in behavioral pain assays after the administration of transporter inhibitors [216, 217]. The analgesic mechanism of glutamate transporter inhibitors may relate to the blockade of inversely operating transporters (i.e., EAATs release rather than uptake glutamate) and the subsequent decrease in release of glutamate into the extracellular space. However, IL-1β, formalin and CFA are used to study acute or inflammatory pain, and, at the time of this review, there are no studies assessing the effects of glutamate transporter inhibitors on NP models.

The sections below will discuss in depth: the regulation of EAATs in animal pain models (3.6), and sex differences in NP and potential roles for EAATs in the maintenance of NP (3.7).

3.4. Neurons and the Transition from Acute to Chronic Pain

All primary sensory neurons secrete glutamate and express glutamate transporters and receptors. In addition, most neurons in the cerebral cortex are glutamatergic excitatory neurons. In the periphery, specialized sensors called nociceptors on glutamatergic neurons detect and convert noxious stimuli into electrical signals, or action potentials, in a process called transduction. Next, transmission occurs as these action potentials travel down peripheral nerve cells classified as Aδ or C fibers that release glutamate, along with other neurotransmitters, at neuronal synapses. These peripheral fibers, or “first-order neurons” are pseudounipolar neurons whose cell bodies congregate in the dorsal root ganglia (DRG) just outside the spinal cord. Aδ fibers, which transmit fast, sharp pain, are myelinated by Schwann cells and have a wider diameter and a faster conduction speed (5–30 m/s) than C fibers. C fibers, however, transmit slow, dull pain, and are unmyelinated. Once they enter the dorsal horn of the spinal cord, they synapse with second-order neurons using glutamate as the primary excitatory neurotransmitter. Many of these nociceptive neurons cross to the contralateral side of the spinal cord and travel up to the brain stem and thalamus where they branch and synapse with third-order neurons to various parts of the brain. These supraspinal locations include the somatosensory cortex, prefrontal cortex, hypothalamus, and various areas of the limbic system (hippocampus, anterior cingulate cortex, insula, amygdala). In addition, second-order neurons branch to the descending modulation pathway, i.e., periaqueductal grey area, rostral ventromedial medulla (RVM), locus coeruleus (LC), etc. to attenuate pain.

There are important molecular changes that occur to neurons in response to persistent pain signals that are still not completely understood. Neuroplasticity, the physical remodeling of neuronal cytoarchitecture, occurs shortly after the onset of persistent acute pain and may contribute to the transition from acute to chronic pain. Changes may initially occur in the periphery with the upregulation of inflammatory mediators that sensitize nociceptors on first-order neurons. Chemicals such as ATP, TNF-α, prostaglandin E2, NGF, bradykinin, serotonin and protons are released by mast cells, macrophages, and epithelial cells to induce peripheral sensitization, which serves to increase pain awareness to limit further damage, fight infection and initiate injury repair. However, ongoing activation of nociceptors changes the expression, trafficking, and distribution of ion channels (e.g., Na_v 1.8 Na⁺ channels) involved in pain transmission. This leads to unstable oscillations of membrane potentials, abnormal firing, and the production of ectopic activity in peripheral afferent neurons. Under normal conditions, noxious stimuli after injury begins to diminish as the healing process occurs until minimal or no pain is detected [218]. Unfortunately for some, the peripheral sensitization and increased nerve firing may affect neurons in the spinal cord and higher structures leading to central sensitization and ongoing pain disorders.

After peripheral nerve injury, neurons in the dorsal horn of the spinal cord display transsynaptic degeneration on the ipsilateral and contralateral sides (although ipsilateral is greater) [219]. Furthermore, peripheral lesions that continuously generate pain impulses to the spinal cord eventually lead to neuronal death in inhibitory interneurons responsible for modulating pain transmission. In rodents, neurons in the dorsal horn were found to undergo apoptosis after nerve injury and these animals subsequently developed hyperalgesia. However, when the NMDAR antagonist MK-801 was administered at the time of injury, there was a decrease in the extent of apoptosis and a decreased onset of hyperalgesia [220]. With ongoing pain signaling, pain transmitting neurons increase in sensitivity and develop more connections to second-order neurons. Eventually, this neuroplasticity leads to central sensitization in the spinal cord and higher CNS centers [218].

Plastic changes are also demonstrated in the descending modulation pathway, such as the RVM, that develops over hours and days after afferent inputs are altered. After injury, neurons in the ipsilateral RVM decreases by over 20% with a subsequent increase in astrocytes and microglia. This loss of RVM neurons decreases pain attenuation and increases pain facilitation [221] and, thus, protecting descending inhibition pathways may prevent the transition from acute to chronic pain [222, 223]. However, failure of a compensatory rebalancing of the RVM output can lead to persistent pain. On and off cells in the descending pathway become sensitized to stimuli but, if normalized or restored, can limit abnormal nociceptive processing [223].

3.5. Glia

Like neurons, glia in the central and peripheral nervous systems express glutamate receptors and transporters and are involved in glutamate regulation and metabolism. Under pathologic conditions, glia disturbances may contribute to chronic pain development [224]. Nerve injury can cause excess release of glutamate that overstimulates postsynaptic receptors leading to increased sensory transmission and neurotoxicity. Extracellular glutamate levels are tightly maintained at low concentrations during physiologic conditions mainly by glial EAATs. Thus, these proteins play a vital role in removing synaptic glutamate and maintaining extracellular concentrations below excitotoxic levels. A brief discussion follows on astrocytes, microglia, oligodendrocytes, satellite glial cells and Schwan cells and their roles in pain. For an overview of subtypes of glia and expression of EAATs, see figure 2.

Figure 2.

Expression of Excitatory amino acid transporters 1–5 (EAAT1–5) on neurons and glia subtypes (astrocytes, microglia, oligodendrocytes, Schwan cells and satellite glia cells (SGCs) in the peripheral and central nervous system. Expression of EAATs may vary in different PNS/CNS regions or during various stages of development.

Created with BioRender.com

3.5.1. Astrocytes

The most abundant cell in the human and rodent nervous systems are star-shaped glial cells called astrocytes. They maintain homeostasis in the CNS by providing structural and nutrient support for neurons and alter neuronal function through extensive contacts with neuronal synapses. Through gap junctions, astrocytes are physically coupled to each other to allow free exchange of ions and cytosolic components [225]. Astrocytes are distinguished from other CNS cell types by the expression of glial fibrillary acidic protein (GFAP), an intermediate filament protein found in astrocytic bodies and processes that increases during astrocyte activation [226]. Additionally, they maintain physiological levels of glutamate by removing 90% of all released glutamate in the CNS using two types of glutamate transporters: Na⁺-independent and Na⁺-dependent transporters[182] (further described in section 3.3. below).

Though rodents are extensively used to study astrocytes, there are distinct differences between rodent and human astrocytes. Human astrocytes contain more subtypes, are twice as large and have ten-times as many primary processes. While a single rodent astrocyte can contact approximate 150,000 synapses, a single human astrocyte can contact up to two million synapses [227, 228]. Thus, the translation from rodent to human subjects in chronic pain that involves the engagement of astrocytes remains challenging.

GLAST/EAAT1 and GLT-1/EAAT2 are the only EAAT subtypes expressed on astrocytes. These EAAT subtypes are localized predominantly on fine astrocytic processes near glutamatergic synapses [190, 229, 230]. EAAT2 represents the majority of EAATs in astrocytes in the brain and spinal cord [190]. Alterations in astrocytic activity is associated with many neurodegenerative and neuropsychiatric disorders including the development of chronic pain [231–233]. During injury or disease, astrocytes change in form and function, a process termed reactive astrogliosis [224, 234, 235]. This activated state results in astrocytic hypertrophy, GFAP upregulation, and altered release of cytokines and gliotransmitters that alter synaptic functions and trigger neuroplasticity [236]. In chronic pain states, astrocytes produce proinflammatory cytokines (e.g., IL-1B, TNF α) that are correlated with increased mechanical allodynia and pain intensity [233]. In addition, astrocytes release neuromodulatory substances, or gliotransmitters, such as d-serine, GABA, ATP and glutamate [237] that also modulates nociception [238].

3.5.2. Microglia

Microglia are a specialized population of resident macrophages in the CNS that are important regulators of brain homeostasis. Through phagocytosis, microglia promptly removes damaged neurons and clears CNS infections. They become activated by many pathologic events and can influence the immune response to a stressor by interacting with infiltrating immune cells though the release of cytokines, chemokines, and growth factors. In addition, there is significant heterogeneity in microglia phenotypes. When chronically activated, microglia may switch from a neuroprotective (M2) to a neurotoxic (M1) phenotype [239], though this is not aways polarized and may occur on a continuum. Microglia have been implicated in neuroinflammation associated with neurogenerative diseases, including SCI, stroke, Parkinson’s disease, Alzheimer’s disease, multiple sclerosis, epilepsy [240], and the generation and maintenance of NP [241–244].

The glutamate transporters GLAST/EAAT1 and GLT-1/EAAT2 are also expressed in microglia. Though the microglial uptake of extracellular glutamate is not as significant as in astrocytes, microglia still play a role regulating glutamate homeostasis after nerve injury. Importantly, while glutamate transporters are often downregulated in astrocytes after injury, the opposite occurs in microglia. Following unilateral facial nerve axotomy in rats, there was an increased expression of GLT-1 in microglia surrounding injured neurons [245]. These results were similar after partial sciatic nerve ligation (pSNL) in rats [246]. At 7–14 days post-injury, animals showed a marked decrease in astrocytic GLT-1 and GLAST expression, however, the microglia had an increased expression of both transporters demonstrating a differential regulation following nerve injury. Despite this increase in microglia, however, animals still demonstrated hypersensitivity to mechanical stimulation suggesting only a modest or absent neuroprotective effect to injury [246].

3.5.3. Oligodendrocytes

Oligodendrocytes are glial cells that myelinate axons in the CNS, however, their role in pain is minimally described and not well understood. Abnormal myelination from oligodendrocyte damage (i.e., SCI, ischemia, multiple sclerosis, ALS) may lead to alterations that contribute to chronic pain. Nonetheless, the limited literature on oligodendrocytes and pain supports a neuroprotective role as they coordinate CNS repair after injury [247, 248]. After SCI, demyelination contributes to the behavioral changes and NP symptoms post-injury. Oligodendrocyte progenitor cells (OPCs) are involved in the remyelination process, and transplantation of these cells after injury promotes functional recovery [249, 250].

The role of glutamate in oligodendrocytes is not well established, but oligodendrocytes do express glutamate transporters and participate in non-synaptic glutamate signaling. These EAATs have the highest expression in oligodendrocytes during early development and may regulate oligodendrocyte maturation, extracellular glutamate homeostasis and CNS myelination [251, 252]. When examining the rat corpus callosum, EAAC1/EAAT3 (rodent/human homologs) on cultured oligodendrocytes contributed the most to glutamate uptake compared to GLAST/EAAT1 or GLT-1/EAAT2. However, there was a high expression of GLT-1/EAAT2 in cultured mature oligodendrocytes compared to in vivo observation (no expression). In addition, unmyelinated axons in the rat corpus callosum release glutamate that elicit AMPAR-mediated currents on glial progenitor cells [253].

During disease, glutamate transporters on oligodendrocytes may become dysregulated. In an animal model of multiple sclerosis, GLAST/EAAT1 and EAAC1/EAAT3 were present in control and MS white matter, however, low levels of GLT-1/EAAT2 expression were observed around MS lesions compared to control animals. Furthermore, glutamate dehydrogenase and glutamate synthetase, enzymes involved in glutamate metabolism, were also downregulated in oligodendrocyte surrounded MS lesions compared to control [254].

3.5.4. Satellite Glial Cells

Glial cells in the PNS regulate the microenvironment surrounding peripheral neurons. In the sensory ganglia, such as the dorsal root, nodose and trigeminal ganglia, neuronal cell bodies are surrounded by satellite glial cells (SGC) [255]. These cells envelope one or several cell bodies and the gap between neurons and SGCs is approximately 20 nm, similar to the width of the synaptic cleft. The morphology of SGCs appears flattened compared to astrocytes, which have branched cytoplasmic extensions. Like stem cells, SGCs contain the transcription factor, Sox2, and may function as progenitor cells. After nerve injury, SGCs demonstrate plasticity, promote regenerative nerve growth, and can promote the growth of myelinating glia and astrocytes [256, 257].

These cells communicate with neurons and each other using chemical mediators such as nitric oxide, ATP, neurotransmitters (i.e., glutamate, substance P), ions (K⁺, Ca²⁺), and immune-mediated cytokines (e.g., TNFα, IL-1β). Using these chemical mediators, SGCs influences the level of neuronal excitability and are involved in the development and maintenance of NP. This occurs through downregulation of K⁺ channels, neuronal hyperreactivity, increased cell-to-cell gap junctions, increased response to ATP, and increased release of proinflammatory cytokines [255, 258, 259].

SGCs are also involved in glutamate modulation using transporters. Early studies using in situ hybridization and immunocytochemistry only found GLAST mRNA and GLAST protein on SGCs [260]. However, more recent groups detected GLAST/EAAT1, GLT-1/EAAT2 and EAAC1/EAAT3 in SGCs of the dorsal root ganglia [261, 262]. It is uncertain whether there is a difference in the expression of glutamate transporters in SGCs during the development of NP.

In cultures of peripheral sensory ganglia, neuronal cell bodies and SGCs both release glutamate and express functional glutamate receptors [263–266]. Glutamate dysregulation leads to increased nociceptive transmission and peripheral sensitization. In the trigeminal ganglia of rats, SGCs expressed GLAST/EAAT1, GLT-1/EAAT2 and glutamate synthase [267]. When glutamate was injected into the ganglia innervating the temporalis or masseter muscle, there was a significant increase in nociception (decrease in afferent mechanical thresholds) that was attenuated by NMDAR antagonists (2R)-amino-5-phosphonovaleric acid, AP-V) or worsened by EAAT blockers (3S)-3-[[3-[[4-(Trifluoromethyl)benzoyl]amino]phenyl]methoxy]-L-aspartic acid, TFB- TBOA) [267].

3.5.5. Schwann Cells

Schwann cells are peripherally located glia that not only myelinate axons but have other non-myelinating phenotypes. One non-myelinating phenotype includes Remak bundles where Schwann cells will embed and bundle several smaller axons together for proper PNS development and regeneration after nerve injury [268]. Non-myelinating Schwann cells can also be found on terminal nerve endings that detect noxious stimuli. Nociceptive neuron terminals are always described as having “free nerve endings” in the epidermis, however, Schwann cells are now found to form a mesh-like network in the subepidermal layers in mouse and humans [269, 270]. In these layers, Schwann cells project extensive processes associated with nociceptive nerves into the epidermis. Experimental ablation of these cutaneous Schwann cells produces nerve retraction, allodynia, and thermal hyperalgesia in mice [270]. In addition, extensive demyelination around peripheral axons also contributes to the development of NP. While their exact role in pain is still poorly understood, the Schwann cells’ proximity to nerves allows for rapid detection and response to peripheral nerve injuries. Like microglia, however, their response varies between proinflammatory and anti-inflammatory phenotypes. In some instances, they act to repair the damaged neuron by releasing glial mediators such as growth factors to support axonal development (e.g., nerve growth factor (NGF), glial cell line-derived neurotrophic factor (GDNF), brain derived neurotrophic factor (BDNF)), anti-inflammatory cytokines (e.g., IL-10) and erythropoietin. During pro-inflammatory states, Schwann cells release the cytokines (e.g., TNFα, IL-1, IL-6) and small molecules (e.g., ATP) that contribute to the initiation of NP.

Although the role of glutamate in Schwann cells is also not well understood, iGluRs and mGluRs have been identified on these cells and some evidence suggests that glutamate may play a role in inducing Schwann cells’ proliferation. For example, AMPAR-mediated currents have been detected during proliferation but when myelination is complete, AMPARs appear to downregulate. In cultured Schwann cells, mRNA was discovered for NMDAR, AMPAR and kainate receptors. When these cells were exposed to glutamate or NMDA, activation of internal cell signaling cascades (Akt, ERK1/2) were also detected[271]. In addition, when glutamate or NMDA was injected directly into crush injured sciatic nerves, ERK1/2 signaling was again observed in both myelinating and nonmyelinating Schwann cells [271]. Furthermore, when mGluR2 is antagonized, the demyelination of injured nerves is inhibited in vivo [272].

3.6. Regulation of EAATs in Animal Models of Pain

Neuronal lesions result in symptoms of neuropathy in humans such as hypersensitivities, spontaneous pain, burning or shock-like sensations, and allodynia. Early preclinical models of pain only focused on acute nociceptive pain using assays such as tail flick or hot plate tests and were not reflective of NP symptoms. A change occurred when a new pain model was created involving surgical injury to peripheral nerves, which generated symptoms similar to peripheral NP conditions [273]. Animal models of pain can be categorized into several groups to include surgical models, drug-induced models, and disease-induced models (see Table 1, which included an overview of pain models). Note that not all models are included since this review focuses on studies that have reported EAATs dysregulation). For a recent review on models for measuring pain in animals, please see [274].

Table 1:

Dysregulation of EAATs in animal models of pain

Surgery, Trauma & Radiation Models
Model	Pain Type	Description	Pros/Cons	EAATs
Chronic constriction injury (CCI) of sciatic nerve	Peripheral NP; some inflammatory pain components; simulates chronic nerve compression and entrapment such as carpal tunnel syndrome [275, 286, 291, 416–418]	Unilateral, mid-thigh incision. 3–4 sutures are loosely tied 1 mm apart around the common sciatic nerve before it branches into common peroneal, tibial, and sural. The sutures are tightened just enough to reduce the diameter of the nerve but not interrupt epineural circulation. This causes some intraneural edema and axotomizes some but not all the nerve axons.	+ : Reliable – consistently produces symptoms of NP that peaks 10–14 days after surgery; easily reproducible; resembles human neuropathy from trauma to peripheral nerves with functional preservation; causes thermal hyperalgesia, mechanical allodynia and signs of spontaneous pain. − : Returns to baseline approximately 20 days after surgery; foreign material (retained sutures) may induce excessive inflammatory reactions; Autotomy (self-mutilation of digits or paws) occurs in some mouse strains; degree of allodynia is less than other models; varying degree of damage based on suture tension	• GLT-1 expression is reduced on postoperative days 7 and 14, along with increased concentration of glutamate in the lumbar spinal cord [288] • Gliosis, reduced GLT-1 expression postoperative days 7 and 14, increased glutamate concentration in the spinal cord dorsal horn [291] • EAAT2 and 3 were downregulated 4 days post-surgery in the spinal cord of rats with hypo HPA-axis reactivity, but no alteration in hyper-reactivity rats. After 21 days, decreased expression of EAAT2 and 3 in hypo and increased EAAT2 in hyper [292]
Spinal Nerve Ligation (SNL)	Peripheral NP	Tight ligation of L5, L5-L6, or L7 spinal nerve distal to the dorsal root ganglia. L4 ligation causes severe motor deficits and must not be damaged or responses to allodynia will not occur. A single L5 ligation is an easier procedure than L5-L6 [293, 419, 420].	+ : Thermal hypersensitivity and mechanical allodynia develops quickly and lasts 4 months; separate injured and intact segments; autotomy is absent but other signs of spontaneous pain are seen (guarding, licking, lifting of paw); the amount of damage is more consistent between animals compared to CCI so there is less experimental variability − : Extensive & difficult surgery; a portion of the paraspinal muscle may be removed producing extra trauma and blood loss[421]; may require special equipment (high-performance stereomicroscope)	• GLT-1 expression goes through an initial upregulation followed by a downregulation [294] • Decreased spinal GLT-1 expression starting at 3 days post-surgery and up to 21 days [295] • EAAT2 is decreased from day 3 to 10 post surgery [100] • Decreased levels of GLT-1 and increased concentration of glutamate at 28 days after SNL [297]. • EAAT2 downregulation in spinal cord 10 days after surgery, prevented by HDAC inhibition [299]. • Decreased GLT-1 expression 7 days after SNL, reverted by administration of inhibitor of JAK/STAT3 [300]
Partial Sciatic Nerve Ligation (pSNL; Seltzer Model)	Peripheral NP; simulates a nerve contusion rather than nerve compression	Tight ligation of one-third to one-half thickness of the sciatic nerve before branching occurs [301, 422]	+ : Produces allodynia and thermal hyperalgesia within hours after surgery that lasts > 7 months; signs of spontaneous pain (guarding, licking); highly reproducible and easier surgery compared to SNL +/− : damaged and undamaged afferent tracts are contained in the same nerve − : Autotomy may occur; number of fibers ligated and damaged may vary between animals	• Downregulation of EAAT2 and EAAT1 levels by 7- and 14-days following surgery. [246]  • GLT-1 expression and uptake activity unaltered at 21 days after injury but increased in hSOD1 G93A rats [303].  • Reduced expression of spinal EAAT1 and EAAT2 and increased EAAT3 at days 4 and 7 post surgery [302] • GLT-1 expression increased on day 3 but reduced on day 10 [327]
Spared Nerve Injury (SNI)	Peripheral NP; studies the neurophysiologic effects of damaged neurons next to intact neurons.	Severe/axotomy of two branches of the sciatic nerve, the tibial and common peroneal nerve; the sural is spared. Care must be taken to not stretch or damage the sural nerve during manipulation and dissection. Other variations include a tibial and sural nerve transection or ligation of only the common peroneal nerve [273, 280, 423, 424]	+ : Less severe motor impairment than CCI; allodynia occurs 2–3 days after SNI and lasts approximately 4 weeks in mice and 15 months in rats; highly reproducible and easy to perform by inexperienced experimenters − : Motor paralysis for several days after surgery but recovers quickly; thermal hyperalgesia may be inconsistently produced or absent; behavioral testing is more difficult as the experimenter must test the area of the paw innervated by the sural nerve only (lateral paw) since the other areas are denervated.	• Initial upregulation of EAAT1 and EAAT2 1 day after surgery, followed by downregulation on postoperative days 3 and 5 [304] • Reactive gliosis, decreased levels of GLT-1 and increased levels of EAAT3 7 days after surgery [305] • Decreased expression of EAAT2 and increased levels of EAAT3 7 days after surgery [307] • Increased mEPSCs frequency and glutamate concentration 4 weeks after surgery, expression of EAATs unchanged [306] • Decreased expression of EAAT1 and EAAT2, but increased expression of EAAT3 28 days after SNI [308] • Downregulation of GLAST and GLT-1 8-, 14- and 20-days post-surgery [412] • Downregulation of GLT-1 in the rat spinal cord 7 and 14 days after surgery, which seems to be regulated by norepinephrine [309] • SNI in GFAP KO and WT mice induced neuropathic behavior and reduction of GLT-1 and GLAST expression after nerve injury, EAAC1 was increased 14 days after SNI [310]
Partial sciatic nerve-transection (PST)	Painful neuropathy from partial nerve injuries	Simple, unilateral, partial sciatic nerve transection without application of foreign material (i.e. sutures). Studies endoneurial changes vs. epineurial inflammation seen in CCI. Less macrophages are seen in this model [281, 425]	+: PST is easy to perform in rats and mice and leads to reproducible pain-related behavior for 5 weeks or longer; rats develop thermal hyperalgesia and mechanical allodynia similar to CCI -: None noted in literature	• Downregulation of GLT-1 and GLAST 1-day post-PST that persists for 21 days, upregulation of EAAC1 in the first 4 days [311] • Down regulation of GLT-1 at day 14, reversed by clavulanic acid administration [390]
Painful nerve root compression (NRC)	Radiculopathy associated with spinal stenosis or disc herniation	Created by using a silicone tube to compress the L5 nerve root [312] Cervical nerve root compression models can also be created by applying microvascular clips to the C7 nerve root [426]	+/− : Produces a recoverable injury. The myelin sheath is lost at day 12 but partially recovers at day 45 and is completely recovered at 5 months [312]	GLAST and GLT-1 expressions decreased in the spinal cord 14 days after NRC [313]
Contusive SCI and other traumatic/ischemic SCI models	Central NP from spinal cord injury [427, 428]	Models include: • Dorsal column crush surgery • SC complete transection between T9-T10 vertebrae • Weight drop (Allen’s model) • Spinal hemisection via laminectomy of T5-T6, T9, or T11-T12 • Clip or forceps compression [429]	+ : Tactile allodynia & thermal hyperalgesia are present although different models of SCI illicit varying degrees of sensory and motor dysfunction. − : Some models have motor impairment; differences between rat and mouse SCI models [430]: mouse models rapidly regenerate after spinal cord transection, which is not seen in rats or humans	• Dysregulation of extracellular glutamate homeostasis and chronic astrocyte activation play roles in persistent hyperexcitability of superficial dorsal horn neurons, leading to maladaptive circuitry, aberrant pain processing and NP [177, 314, 315, 317, 318]
Drug-Induced Models
Model	Conditions Studied	Description	Pros/Cons	EAATs
Formalin Injection	Inflammatory pain, orofacial pain, acute pain, neuropathic-like pain symptoms (depending on concentration)	Formalin, a colorless solution of formaldehyde, is injected into rodent paws, temporomandibular joints and maxillae [319, 320, 431]	+ : Formalin injection produces concentration-dependent mechanical allodynia and hyperalgesia for approximately 14 days similar to peripheral nerve injury − : Several drugs (gabapentin, morphine) may reverse 1% formalin injection but not 2% or 5%; may not be reflective of all NP conditions and has a large inflammatory pain component	• Formalin induces nociceptive behavior and acute GLT-1 downregulation in the hippocampus and ACC, reverted by LDN-212320 administration [321]
Complete Freund’s Adjuvant (CFA)	Inflammatory pain	CFA is a suspension of mycobacteria that can be injected subcutaneously into a paw or intraperitoneally to induced inflammatory pain [322, 432]	+ : Only takes 24 hours to have a peak effect − : Lasts about 7 days and requires repeated injections for long-term studies	• Downregulation of GLAST and GLT-1 2 and 3 days after CFA-injection [411] • After CFA -induced hind paw inflammation, GLT-1 and KBBP which were upregulated early, which transitioned to downregulation for 8 weeks., consistent with hyperalgesia [323] • Downregulation of GLT-1 expression at 1, 3, and 7 days but not 14 days, in rat spinal dorsal horn, associated with increases in miRNA-107 [282]
Chemotherapeutic agents or Cancer-related NP as a consequence of cancer-directed therapies	Chemotherapy-induced peripheral neuropathy (CIPN) from vinka alkaloids, platinum compounds, proteasome inhibitors and taxols [433]	Models include: • Vincristine-induced peripheral neuropathy: repeated injections produces a dose-dependent mechanical allodynia • Paclitaxel-induced peripheral neuropathy: a single injection induces neuropathy within 24–72 hours • Cisplatin-induced peripheral neuropathy: repeated daily injections of Oxaliplatin [203, 434, 435]	+ : Mimics peripheral neuropathies seen in patients; large fiber deformities such as axonal degeneration and demyelination may last up to 4 months − : Inconsistent dose, mode of delivery, duration or sex, age and genetic background of rodent causes varying degrees of allodynia; IP injections of drugs do not model human delivery; most behavioral assays are evoked responses, however, most patients with CIPN experience spontaneous pain, which is difficult to assess in rodents; research is mostly conducted on rats	• Downregulation of GLAST through 16 days after paclitaxel treatment, and GLT-1 at 4 hours after treatment that recovered at 7 days after treatment [325] • Decreased expression of GLT-1, 10 days after the first taxol injection, reverted by a GSK3β inhibitor [326] • Increased levels of GFAP immunoreactivity 7 days after paclitaxel administration, but no alteration in EAATs levels [328] • Oxaliplatin-treated rats have elevated baseline glutamate concentration and decreased expression of EAAT2 in the spinal cord, reverted by riluzole [329] • Paclitaxel administration in rats increased glutamate concentration and decreased EAAT2 expression 21–28 days later [203]
Opioid-induced hyperalgesia (OIH)	Opioid-induced hyperalgesia (e.g. remifentanil, fentanyl, morphine)	Models can be produced by repeated boluses (2x/day) or continuous administration of opioids via catheter either IP, intrathecally or subcuteanously [436, 437] Murine OIH models can also be created with subcutaneously implanted opioid pellets [438]	+ : mechanical allodynia and thermal hyperalgesia is produced similar to human OIH − : OIH models are not consistent or standardized between studies making it difficult to compare results	• Reduced spinal GLT-1 expression, reverted by ceftriaxone [330] • Decreased GLT-1 expression induced by chronic morphine, the effect is at least in part associated with spinal astrocytic Cx43 [439] • Decreased GLT-1 expression in spinal cord 48 h after remifentanil infusion [331].
Disease-Induced Models
Model	Conditions Studied	Description	Pros/Cons	EAATs
Diabetes Models	Diabetic peripheral neuropathy	Models include: Zucker diabetic fatty (ZDF) or leptin receptor decrease (db/db): missense mutation in a gene encoding the leptin receptor that causes hyperglycemia, insulin resistance, hypertension and obesity, similar to type-2 diabetes [332] • Streptozotocin (STZ)-induced models: STZ is an antibiotic that destroys insulin-secreting pancreatic cells (Islet of Langerhans). Produces hyperglycemia [440] • Non-obese diabetic mice: High-glucose, high-fat diet mice • Leptin (ob/ob): develop polyphagia, obesity, insulin resistance and Type 2 diabetes mellites [441] • Transgenic models: TCR-SFE/Ins-HA mouse (double transgenic mouse of T1DM) [441] or genetic models causing autoimmune damage to the pancreas or misfolded insulin [442]	+ : some models can produce long-lasting thermal and mechanical hyperalgesia and allodynia − : the severity of induced diabetes can vary significantly between studies; models can produce severe distress and deterioration making it difficult to do behavior data and collect pain scores; pathologic features of advanced diabetic neuropathy (i.e. demyelination and degeneration) are rarely seen in rodent models [442]; the phenotype of diabetic peripheral neuropathy models may resemble human symptoms, however, the etiology of disease may not correlate limiting the translation of therapies to humans	• ZDF rats have higher levels of extracellular glutamate and decreased levels of GLT-1 expression in the dorsal horn of the spinal cord, reverted by ceftriaxone [333] • STZ-induced NP model in rats result in decreased GLT-1 expression in the spinal cord 21 days later, which is restored by clavulanic acid administration [335]
Multiple Sclerosis Model	Multiple sclerosis-induced NP	Models include: • Experimental autoimmune encephalitis (EAE): mouse immunization with myelin oligodendrocyte glycoprotein. Widespread CNS inflammation, denervation and locomotion impairment occurs. Tactile allodynia occurs before overt neurological deficits • Theiler’s murine encephalomyelitis virus (TMEV): mouse infected with TMEV (intracerebral inoculation) causes mechanical allodynia and thermal hyperalgesia [443–445]	+ : EAE is a well-characterized model of MS that produces pain sensitivity, which can be assessed by well-established assays [446] − : concurrent severe motor impairments through the disease time course interferes with assessment of pain behaviors	• Reduced expression of GLT-1 in lumbar spinal cord, reverted by ceftriaxone [382] • Decreased spinal levels of EAAT2, also reverted with ceftriaxone [337]
Visceral Pain	Visceral pain; pain arising from damage to internal organs, particularly the gastrointestinal tract (GIT)	Colorectal distension (CRD) using a balloon in the colorectal cavity. A barostat is used to monitor pressures. This has replaced previous methods of injecting irritant chemicals into the GIT. Animal vital signs (heart rate, respiration and blood pressure) or a visceromotor (electromyographic recordings of abdominal muscles) can be used to measure noxious intensities [338].	+ : Physical distension of the colon is thought to replicate human visceral pain and similar patterns of referred pain [447]; CRD is restricted to the viscera itself − : No reported downsides to CRD. Colonic instillation of irritant chemicals is poorly reproducible, creates long-lasting pain and may not be related to human pathology [338]	• 7 days administration of ceftriaxone mitigates visceral nociception [296]. • Intrathecal delivery of DHK reversed the effect of ceftriaxone [296].

Various animal pain models have been utilized to study the regulation of EAATs during the development of NP. Several pre-clinical studies have reported that EAATs undergo differential, temporal regulation in NP states. However, these findings are inconsistent, which may result from the use of different models, variabilities in types and times of tissue collection, and approaches used for EAAT expression assessment (such as western blots vs. qRT-PCR).

3.6.1. Surgery & Trauma Models

The earliest models of NP with high validity have utilized the surgical manipulation of peripheral or central neurons [275, 276]. Surgical models of NP typically involve partial denervation through constricting or ligating a branch of a peripheral nerve, such as the sciatic nerve, or damaging a portion of the spinal cord to simulate human pain conditions. Surgical or trauma models used to study EAATs include chronic constriction injury (CCI) [277], spinal nerve ligation (SNL)[278], partial sciatic nerve ligation (pSNL)[279], spared nerve injury (SNI) [273, 280], partial sciatic nerve transection (PST)[281], nerve root compression (NRC)[282] and SCI [275, 283]. Most of these models reliably produce long-lasting symptoms of allodynia and hypersensitivity which can be tested using various behavioral assays [280, 284, 285].

3.6.1.1. Chronic Constriction Injury (CCI)

Unilateral chronic constriction injury (CCI) is a widely used animal model developed by Bennett & Xie [286] that involves the application of several sutures loosely tied around the sciatic nerve to occlude but not arrest the neuronal blood flow. Animals that undergo CCI show hyperalgesia, allodynia, and dysesthesia (spontaneous pain) similar to human chronic NP. Before CCI was used to study EAAT regulation, researchers observed a lack of glutamate uptake in the spinal cord after partial and complete nerve injury [287]. Subsequently, Sung et al. found that the expression and uptake activity of spinal EAATs were altered in the rodent CCI model. Intriguinly, CCI induced an initial up upregulation of EAATs in the ipsilateral spinal cord dorsal horn 5 days after the injury, followed by downregulation on postoperative days 7 and 14. This study was the first to suggest that changes in expression and activity of spinal EAATs could play a role in both the induction and maintenance of NP[288]. The role of EAAT modulation in pain was further supported in rodent studies demonstrating that the administration of the EAAT inhibitors DL-TBOA, DHK, and PDC resulted in a pain phenotype[153, 289].

Hu et al., however, used the CCI model to show a downregulation of spinal EAATs. CCI induced significant thermal hyperalgesia and mechanical allodynia from postoperative day 3 to day 21, which was accompanied by significant downregulation of GLT-1 expression in the spinal cord. Further, they determined that administration of ceftriaxone, a β-lactam antibiotic that was previously characterized to increase transcription expression of GLT-1, upregulated GLT-1 expression and glutamate uptake in the spinal dorsal horn with anti-nociceptive effects[290].

CCI was also used to examine reactive changes in glial cells after injury, termed gliosis, in the dorsal horn of the spinal cord. Three days after CCI, there was an increase in cytokines, proinflammatory mediators and microglia, which decreased on postoperative day 7 [291]. On day 7, levels of spinal GFAP were also increased, suggesting gliosis, which lasts for at least 14 days and beyond. The expression of glial glycine and glutamate transporters (GlyT1 and GLT-1, respectively) was reduced on postoperative days 7 and 14, along with increased concentration of glutamate and glycine from the lumbar spinal cord of CCI mice. This study suggested that microglial activation precedes astrogliosis, and astrogliosis correlates with a marked decrease in glycine and glutamate transporters resulting in increased concentration of these neurotransmitters in the spinal cord. As a result, gliosis was strongly implicated in the development and maintenance of persistent pain states following CCI.

The CCI model was used to investigate the onset and maintenance of mechanical allodynia/hyperalgesia and the expression of biochemical mediators involved in spinal cell modulation. Two rat strains were used that displayed either hypo- (Lewis-LEW) or hyper- (Fischer 344-FIS) reactivity of the hypothalamus-pituitary-adrenal (HPA) axis. 4 days after CCI surgery, EAAT2 and 3 were downregulated in the spinal cord of LEW rats but no EAAT alterations occurred in FIS and Wistar rats. After 21 days, there was decreased expression of glutamate transporters in LEW, but an increased expression of EAAT2 in FIS. This finding is inconsistent with previous work demonstrating that CCI or inflammation causes a rapid transient upregulation of EAATs in the spinal cord, followed by a downregulation [288]. Overall, the interaction between the HPA axis and the spinal activation pattern is complex in NP, and causal relationships cannot be determined between pain behaviors, glial cell activation or the expression of the glutamate transporters without further studies [292].

3.6.1.2. Spinal Nerve Ligation (SNL)

Spinal nerve ligation (SNL) is a animal pain model produced by tightly ligating the L5 or L5-L6 spinal nerves just distal to the dorsal root ganglia [293]. A study with the SNL model revealed temporal changes of astrocyte activation and GLT-1 expression in the spinal cord [294]. In accordance with early studies, the expression of GLT-1 displayed a biphasic change, with an initial upregulation followed by downregulation, and a long-term (> 21 days) activation of astrocytes in the ipsilateral spinal dorsal horn. In accordance with these results, another study showed a significant reduction in the expression of GLT-1 and GLAST in the spinal dorsal horn. Mechanical withdrawal threshold were also reduced in this model from days 3–10 post-surgery, however, a biphasic behavior was not observed, probably because early and later time points were not measured. In the time points studied, administration of VPA reduced hypersensitivity and restored the expression of GLT-1 and GLAST in the spinal dorsal horn [100].

Further studies by Maeda et al. transferred the GLT-1 gene recombinant adenoviruses into the rat spinal cord, which resulted in increased GLT-1 expression on astrocytes at 2–21 days after viral administration. The spinal GLT-1 gene transfer did not affect acute mechanical and thermal nociceptive responses in naive rats. However, it reduced inflammatory mechanical hyperalgesia in an inflammatory model, and attenuated the decrease in spinal GLT-1 expression in the SNL model at 3 days and up to 21 days post-SNL. The study did not observe an early upregulation of GLT-1 that was reported in previous studies. GLT-1 gene transfer 7 days before SNL also prevented the induction of tactile allodynia, but not when delivered 7 or 14 days after SNL. This suggests that overexpression of GLT-1 on spinal cord astrocytes attenuates the induction, but not the maintenance, of inflammatory pain and NP. This may occur by preventing the induction of central sensitization without affecting acute pain sensation [295]. A subsequent study in a mouse model of visceral pain (colon distention) also used viral gene transfer of GLT-1 and found that upregulation of GLT-1 led to a significant reduction of pain behaviors. However, the levels of GLT-1 before viral gene transfer were not investigated in this pain model [296].

The regulation of GLT-1 and mGluRs in the LC was investigated using a model of subdermal injection of capsaicin that induces LC neuronal activity in normal and SNL rats. SNL resulted in decreased levels of GLT-1 and increased concentration of basal extracellular glutamate at 28 days after SNL. Additionally, capsaicin-evoked LC neuronal activation was decreased in SNL rats, and removal of autoinhibition of glutamatergic terminals by inhibition of mGluR or increasing GLT-1 expression by inhibition of HDAC (histone deacetylase) restored noxious stimulus-induced analgesia in SNL rats. Finally, knockdown of GLT-1 in the LC mimicked SNL-induced impairment of noxious stimulus-induced analgesia in normal rats, demonstrating the importance of GLT-1 regulation in the LC in NP [297]. Another study examined the role of GLT-1 in the LC in the context of gabapentin administration to understand why gabapentin sometimes lacks efficacy in NP. Previous studies demonstrated that gabapentin is efficacious only within the first few weeks of SNL in rats. Administration of gabapentin had a time-dependent reduction in anti-hypersensitivity after SNL associated with downregulation of GLT-1 in the LC. SNL also resulted in a time-dependent increase in basal noradrenergic neuronal activity in the LC that was reversible with gabapentin. Knock-down of GLT-1 in the LC reduced the anti-hypersensitivity effect of gabapentin 2 weeks after SNL. On the other hand, increasing GLT-1 expression with the HDAC inhibitor, valproate, restored the anti-hypersensitivity effect of gabapentin 8 weeks after SNL. Knock-down of GLT-1 in the LC reversed the effect of valproate to restore gabapentin-induced anti-hypersensitivity. Downregulation of GLT-1 in the LC and reduced spinal noradrenergic inhibition contribute to reduced analgesic efficacy from gabapentin in chronic NP, and valproate can rescue this impaired efficacy [298]. A consequence of HDAC activation is downregulation of astrocytic glutamate transporters in the spinal cord, resulting in elevation of extracellular glutamate level and contributing, at least in part, to development of chronic pain after nerve injury.

When SNL model animals were administered the compound suberoylanilide hydroxamic acid (SAHA), a HDAC inhibitor, there was a reversal in the downregulation of spinal EAAT2 and an attenuation of pain behaviors. SAHA was intrathecally injected in rats 5 days to 7 days after SNL, resulting in significant attenuation of mechanical allodynia and thermal hyperalgesia, which was correlated with significant prevention of SNL-induced downregulation of EAAT2 in the spinal dorsal horn 10 days after the surgery. Overall, HDAC may contribute to the downregulation of EAAT2, and this mechanism should be considered for the development of more effective strategies for treating NP [299].

Astrocytic N-myc downstream-regulated gene 2 (NDRG2), a tumor suppressor protein, and the stress response was investigated in the pathogenesis of NP in the SNL model. It was found that NDRG2, which is mainly expressed in astrocytes, was increased in the spinal cord after SNL in rats. Furthermore, suppression of NDRG2 alleviated SNL-induced mechanical and thermal hypersensitivity, as well as elevated GLT-1 expression and downregulated pro-inflammatory cytokine levels in the spinal dorsal horn. Experiments in primary astrocyte cultures stimulated with lipopolysaccharide (LPS) showed that inhibition of NDRG2 reversed both the LPS-induced activation of astrocytes and decreased expression of GLT-1. However, overexpression of NDRG2 resulted in activation of astrocytes, aberrant glutamatergic neurotransmission, and spontaneous nociceptive responses in rats. In addition, administration of a JAK/STAT3 (Janus kinase-signal transducer and activator of transcription 3) signaling pathway inhibitor attenuated mechanical and thermal hyperalgesia, inhibited reactive astrocytes, and restored normal expression levels of astrocytic GLT-1 in the spinal dorsal horn of NDRG2-overexpression rats. Spinal astrocytic NDRG2 were found to be critical for the maintenance of NP through modulation of astrocytic and inflammatory responses via regulating GLT-1 expression through the JAK/STAT3 signaling pathway, suggesting NDRG2 as a novel therapeutic target for NP treatment of NP [300].

These studies support both astrocyte activation and GLT-1 expression changes occurs in NP and may contribute to the initiation and/or maintenance of NP.

3.6.1.3. Partial Sciatic Nerve Ligation (pSNL)

Another animal model of peripheral NP is the partial sciatic nerve ligation (pSNL) model, where a suture is inserted through one-third to one-half of the common sciatic nerve and tightly ligated so only a portion of the nerve is injured [301]. A significant downregulation of total EAAT1 and EAAT2 levels was observed 7 and 14 days after pSNL in rats, as well as hypersensitivity to mechanical stimulation [246]. Microglia and astrocytes were reactive at 7 and 14 days following pSNL, measured by increased expression of OX-42 and GFAP. Additionally, pSNL seemed to have altered the cellular location of EAAT1 and EAAT2 transporters, in that activated astrocytes had a marked decrease in their expression, whereas activated microglia showed de novo expression of GLT-1 and GLAST at 7 days after pSNL through day 14. This study indicated that the expression of glutamate transporters in astrocytes and microglia are differentially regulated following nerve injury. No upregulation was observed in this study, but time points earlier than 7 days were not examined. In accordance, another study showed decreased expression of spinal EAAT1 and EAAT2 at 4 and 7 days after pSNL, whereas neuronal glutamate transporter EAAT3 was increased, suggesting a compensatory mechanism, and providing more evidence that astrocytic transporters dysfunction is associated with nerve injury-induced pain [302].

The glial activation and inflammatory processes in the pSNL model were studied using transgenic rats expressing mutated superoxide dismutase 1 (hSOD1 G93A). pSNL induced thermal and mechanical hypersensitivity in both WT and transgenic rats, and exacerbated hypersensitivity in transgenic rats. Microglial activation was increased in the ipsilateral dorsal horn of the lumbar spinal cord after pSNL, which was further enhanced in the transgenic rats. On the other hand, the pSNL-induced increase in GFAP immunoreactivity observed in WT rats was unexpectedly found to be attenuated in transgenic pSNL rats. GLT-1 gene expression and uptake activity were shown to be similar between WT sham and WT ligated rats at 21 days after injury, however, early time points were not investigated decreasing comparability with other studies. Finally, in transgenic rats, GLT-1 expression and function were significantly increased in the ipsilateral dorsal region of the lumbar spinal cord [303], which seemed surprising, however, this could be associated to differential expression in astroglia and microglia, as it was previously reported after pSNL in the above study[246].

3.6.1.4. Spared Nerve Injury (SNI)

Spared nerve injury (SNI) is a peripheral NP model produced by severing the common peroneal and tibial branches of the sciatic nerve while leaving the sural nerve intact. An early study in this model found that expression of EAATs has a biphasic pattern after injury [304]. There was an initial upregulation of EAAT1 and EAAT2 on postoperative day 1 in the ipsilateral dorsal horn of the spinal cord, followed by downregulation on postoperative days 3 and 5. Additionally, amitriptyline reversed the downregulation of EAATs on postoperative days 3–5 and attenuated the symptoms of mechanical allodynia.

Analysis of spinal cord sections in the rat SNI model 7 days after surgery revealed increased microglia and astrocytic markers, which indicated reactive gliosis. This study also observed decreased levels of glial transporters GLT-1 and GlyT1, and increased levels of EAAT3. In addition, other changes in the GABA systems resulted in an increased glutamate/GABA ratio, which supports an association between reactive astrogliosis and mechanisms underlying the perturbation of the synaptic circuitry in the SNI model [305].

The SNI model was used to record miniature excitatory postsynaptic currents (mEPSCs) from neurons of the rat dorsal horn to evaluate spontaneous synaptic activity. These investigators observed a doubling of the frequency of mEPSCs after SNI, which suggests augmented release of glutamate from primary afferents or spinal interneurons. The concentration of glutamate in the cerebrospinal fluid was also elevated at 1 and 4 weeks after SNI, however, the expression of EAATs remained unchanged, which is different from previous studies that observed biphasic EAAT regulation or downregulation. An antagonist of mGluR5 reduced the frequency of mEPSCs to control levels, suggesting a positive feedback mechanism that involves the facilitation of glutamate release through activation of mGluR5 by high extracellular glutamate. Treatment with ceftriaxone increased GLT-1 expression in the dorsal horn after SNI, resulting in increased glutamate uptake and lower extracellular glutamate, which prevented the facilitation of transmitter release by mGluR5 and attenuated NP-like behavior. Consequently, balancing glutamate release and uptake after nerve injury is an important target for the management of chronic NP [306].

BB14, a nerve growth factor-like peptide, was reported to reduce reactive astrogliosis and restoring synaptic homeostasis in the SNI rat model. There was also a decreased expression of GLT-1, increased levels of EAAT3, increased glutamate/GABA ratios and a reduction of glutathione levels at 7 days after SNI in the dorsal horn of the lumbar spinal cords. Biochemical alterations and SNI-related neuropathic behavior, characterized by allodynia and hyperalgesia, were reversed by administration of BB14. This supports a correlation between reactive astrogliosis and disruption of synaptic circuitries in SNI and supports a potential role of BB14 as a therapeutic agent [307].

Modulation of reactive gliosis in the spinal cord after SNI has also been investigated as a strategy to restore synaptic homeostasis. In mice, there was increased glial response, decreased expression of EAAT1 and EAAT2, but increased expression of EAAT3 at 28 days after SNI, suggesting compensatory mechanisms are involved in the regulation of these EAATs. Spinal neurons and astrocytes showed a persistent increase of calcium levels. This study supported a correlation between reactive gliosis and disturbance of the spinal synaptic homeostasis following SNI [308].

A study using the SNI model reported a bilateral downregulation of GLT-1 in the rat spinal cord 7 and 14 days after SNI. Seeking an association between GLT-1 regulation and norepinephrine, the authors demonstrated that norepinephrine induced downregulation of GLT-1 in primary astrocytes from rat spinal cords. Furthermore, the effect of norepinephrine was reversed by an α-adrenoceptor antagonist [309]. However, further studies are needed to better understand the mechanisms associated with GLT-1 downregulation by norepinephrine

The maladaptive response of the CNS in the SNI model was also investigated, which is attributed to activation of glial cells and subsequent inflammatory reaction and a morphological and functional remodeling. GFAP was measured to evaluate reactive astrocytes after SNI in GFAP KO and WT mice. SNI induced neuropathic behavior in both GFAP KO and WT animals, which was associated with intense microglial reaction, reactive astrocytosis and alteration of glial and neuronal glutamate and GABA transporters. Specifically, expression of GLT-1 and GLAST were reduced after nerve injury whereas EAAC1 was increased 14 days after SNI [310].

3.6.1.5. Partial Sciatic Nerve Transection (PST)

Partial sciatic nerve transection (PST) involves incompletely severing the sciatic nerve as an alternative to CCI models that introduce foreign material (i.e. sutures), which causes inflammatory reactions [281]. One study using the PST model in rats revealed downregulation of GLT-1 and GLAST in the spinal cord as early as 1 days post-PST and persisted for at least 21 days. Conversely, EAAC1 upregulation was observed in the first 4 days after surgery [311]. These investigators found that intrathecal administration of ultra-low dose of naloxone enhanced the antinociceptive effect of morphine by increasing glutamate uptake in the spinal cord of the PST rats. These findings conflict with the studies that observed a biphasic regulation of GLT-1 after nerve injury.

3.6.1.6. Nerve Root Compression (NRC)

Nerve root compression (NRC) is a model of low-back pain syndromes and sciatica and is created by compressing the L5 nerve root with a silicone tube [312]. In NRC, expressions of GLAST, GLT-1, mGluR5, and GluR1 are decreased in the spinal cord 14 days post procedure. Inhibition of spinal secretory phospholipase A2 (sPLA2) attenuated pain and spinal neuronal hyperexcitability through alterations of spinal glutamatergic signaling and restored GLT-1 expression [313].

3.6.1.7. Spinal Cord Injury (SCI)

Development of NP also occurs in traumatic SCI patients resulting in debilitating and long-term physical and psychological burdens. Following SCI, chronic dysregulation of extracellular glutamate homeostasis and glutamate transporter function and expression are shown to play a key role in persistent central hyperexcitability of superficial dorsal horn neurons. Consequently, this mediates pain neurotransmission leading to various forms of NP [177, 310, 314–318].

Several studies from the Lepore laboratory investigated the role of spinal EAATs in rat models of cervical contusion SCI. Contusion trauma to cervical spinal cord is a form of pathological nociception that occurs in a significant portion of traumatic SCI patients, resulting in debilitating and often long-term physical and psychological burdens. Following SCI, chronic dysregulation of extracellular glutamate homeostasis is shown to play a key role in persistent central hyperexcitability of superficial dorsal horn neurons that modulate pain transmission [314]. Spatial and temporal changes in promoter activity of GLT-1 following traumatic SCI were also observed at the lesion epicenter and at rostral and caudal areas where no tissue loss occurred, suggesting that both astrocyte death and gene expression changes in surviving astrocytes compromises glutamate homeostasis [318]. One study determined that chronic thermal hyperalgesia was accompanied by neuronal and astrocyte activation and loss of GLT-1 2 and 6 weeks after SCI in the superficial dorsal horn [177]. A follow-up study reported that this model induces persistent thermal hyperalgesia, tactile allodynia, and chronic neuronal and astrocyte activation in the dorsal horn. They investigated the association between pain behavior and glutamate transporters levels and observed a decrease in GLT-1 levels the ipsilateral dorsal horn of cervical spinal cord. Additionally, robust hyperalgesia in the ipsilateral forepaw and allodynia in both forepaws were evident within two weeks following injury and persisted for at least six weeks [317]. This study suggests that activation and proliferation of astrocytes as well as loss of GLT-1 function contribute to central sensitization and underlie the development of NP after SCI.

The implications of GLT-1 reductions were further investigated in the cervical spinal cord dorsal horn after unilateral cervical contusion SCI. They used adeno-associated virus type 8 (AAV8)-Gfa2 vector injected into the superficial dorsal horn to increase GLT-1 expression selectively in astrocytes. In the injured cervical spinal cord, AAV8-GLT-1 delivery increased GLT-1 protein expression in astrocytes of the dorsal horn, resulted in persistent reversal of heat hypersensitivity, and significantly reduced the expression of ΔFosB, a transcription factor and marker of persistently increased neuronal activation. This study supports that focal restoration of GLT-1 expression in the superficial dorsal horn is a promising target for treating chronic NP following SCI [315]. In conclusion, SCI can induce cellular changes in the CNS, cause central sensitization, and alter the excitability of spinal cord neurons, especially those in the dorsal horn involved in pain transmission.

3.6.2. Drug-Induced Models

Commonly used drug-induced models can be used to simulate certain conditions such as inflammatory pain, or pain produced as an adverse effect of toxic medications such as chemotherapy or anti-retroviral drugs. Formalin or Complete Freund’s adjuvant (CFA) injection can be used to quickly induce inflammatory pain and can easily be used to study anti-nociceptive compounds during drug discovery. Chemotherapeutic drugs such as vincristine, cisplatin and taxanes (taxols), may be used to study peripheral neuropathy, a common side effect of chemotherapy [276].

3.6.2.1. Acute models

Formalin

Formalin injection into temporomandibular joints and maxilla induces acute orofacial inflammatory pain[319, 320], nociceptive behavior and acute GLT-1 downregulation in the hippocampus and anterior cingulate cortex (ACC), important areas of pain regulation. The nociceptive behavior and GLT-1 downregulation were shown to be reversed by administration of LDN-212320, a compound that increases expression of GLT-1 [321].

Complete Freund’s Adjuvant (CFA)

CFA is a suspension of desiccated mycobacterium that can be injected subcutaneously into a paw or intraperitoneally to induce inflammation and tissue necrosis in approximately 24 hours [322]. A study focused on the RVM, a key site in pain modulation, examined the expression and function of GLT-1 and related transcription factor kappa B-motif binding phosphoprotein (KBBP) in a rodent model of CFA -induced hind paw inflammation. After inflammation, GLT-1 and KBBP had an early upregulation (at 30 min) and gradual downregulation for 8 weeks. Mechanical hyperalgesia and paw edema showed an initial developing phase with peaked hyperalgesia at 4 to 24 h, followed by an attenuating phase and a late persistent-lasting phase. GLT-1 downregulated during the transition into the persistent hyperalgesia phase. In the RVM, pharmacological block with DHK and RNAi of GLT-1 and KBBP increased nociception and overexpression of GLT-1 reversed persistent hyperalgesia. This study suggests that the initial increased GLT-1 activity depends on injury input and may be a compensatory mechanism to inhibit the development of hyperalgesia. Later downregulation of GLT-1 may promote the persistence and maintenance of pain [323].

Another recent study implicated miRNA-107 as a contributor factor in inflammatory pain in the CFA model by down-regulating GLT-1 expression in rat spinal dorsal horn. GLT-1 was significantly decreased at 8 hours, 1, 3 and 7 days after CFA administration, however, at 14 days the GLT-1 levels returned to baseline. They also found that miRNA-107 and GLT-1 were co-expressed in the same cells of the spinal dorsal horn in CFA rats, suggesting that an increase of miR-107 contributes to inflammatory pain through downregulating expression of GLT-1 expression, suggesting the miR-107/GLT-1 signal pathway as a potential strategy for pain therapeutics [282].

3.6.2.2. Chronic models

Chemotherapy-Induced NP (CINP)

An adverse effect from chemotherapeutic drug therapy is the development of temporary or permanent peripheral neuropathy. The contribution of spinal astrocytes EAATs to the pathogenesis of CINP was studied in the paclitaxel-induced NP model, where a rapid and persistent activation of spinal astrocytes was observed with no increase in pro-inflammatory cytokines, but with a significant downregulation of GLAST and GLT-1 in the spinal dorsal horn. The downregulation of GLAST by paclitaxel was rapid and significant, as early as 4 hours after treatment, and remained significantly decreased up to 16 days post-treatment. On the other hand, the downregulation of GLT-1 was only observed at 4 hours after paclitaxel treatment and then recovered at 7 days after treatment. Systemic treatment with minocycline, a selective inhibitor of microglia activation[324], prevented activation of astrocytes and downregulation of GLAST and GLT-1 in the spinal dorsal horn. This may be due to the involvement of spinal astrocytic EAATs rather than microglia in the pathogenesis of paclitaxel-induced neuropathy [325].

In addition, two critical spinal mechanisms underlying paclitaxel-induced NP are an increased production of pro-inflammatory cytokines and decreased activities of EAAT1 and EAAT2. Ten days after taxol initiation, activation of glycogen synthase kinase 3beta (GSK3β) in the spinal dorsal horn was associated with an increased expression of GFAP and IL-1β, a decreased expression of GLT-1, and the development and maintenance of (?) CINP. Administration of pre-emptive lithium (a GSK3β inhibitor) resulted in less mechanical and thermal allodynia, while protein expressions of GLT-1, GFAP and IL-1β in the spinal dorsal horn were increased. This suggests that the suppression of spinal GSK3β is a critical mechanism to reduce taxol-induced CINP [326]. However, when using the pSNL model, there was a dynamic alteration of GSK3β activities that was suppressed on day 3 but increased on day 10 following nerve injury. GLT-1 expression in the spinal dorsal horn was increased on day 3 but was reduced on day 10, in agreement with previous studies that revealed a biphasic change. On days 3 and 10, astrocytes were activated and over-produced IL-1β, and thermal hyperalgesia and mechanical allodynia was observed. Pharmacological inhibition of GSK3β activities ameliorated the pain behaviors, prevented decreased GLT-1 expression, and inhibited astrocytic activation and levels of IL-1β in the spinal dorsal horn on day 10. This study suggests that the increased GSK3β activity in the spinal dorsal horn is due to GLT-1 downregulation at later stages of injury, suggesting GSK3β as a potential target for the development of CINP therapeutics [327].

Glutamatergic neurotransmission was investigated in the paclitaxel-induced CINP in mice. 7 days after administration of paclitaxel, researchers observed thermal hyperalgesia, increased levels of GFAP transcripts and GFAP immunoreactivity, however, levels of GLAST, GLT-1, EAAC1, EAAT4, vGLUT-1 and vGLUT-2 were not significantly altered. They observed a differential upregulation of glutamate receptor subunits in the ACC. This study suggests that targeting astrocyte activation and the glutamatergic system might be another therapeutic avenue for management of paclitaxel-induced CINP [328].

The regulation of spinal EAAT2 and vGLUT2 via HDAC2 was studied in paclitaxel-induced painful neuropathy. In rats, paclitaxel induced mechanical allodynia for longer than 28 days, which was accompanied by increased glutamate concentrations and decreased EAAT2 expression. On the other hand, no changes in GABA and glycine transporters or vGAT (vesicular GABA transporter) in the spinal dorsal horn were observed. HDAC2 and transcription factor Yin Yang 1 (YY1) and vGLUT2 were upregulated, and inhibition or knockdown of HDAC2 expression attenuated mechanical allodynia. This process suppressed HDAC2 upregulation, increased glutamate, and changed EAAT2/vGLUT/synaptophysin expression and the interaction of HDAC2/YY1. Dysregulation between glutamate release and uptake due to dysfunction of the EAAT2/vGLUT2/synaptophysin cascade in the spinal dorsal horn plays a key role in the development of paclitaxel-induced NP. In addition, the interaction of HDAC2 and YY1 potentially regulates this pathway, and this process may be a therapeutic target to relieve CINP by decreasing central sensitization of spinal nociceptive neurons [203].

Oxaliplatin is another chemotherapeutic that causes severe peripheral neuropathy. One study looked at spinal glutamate transmission in oxaliplatin-induced mechanical allodynia and investigated the effects of riluzole. In vivo spinal microdialysis revealed that baseline glutamate concentrations were elevated in oxaliplatin-treated rats, mechanical hind-paw sensitivity increased, and EAAT2 expression was decreased, measure on day 28 after repeated oxaliplatin administration. Subsequent administration of riluzole attenuated the increase in glutamate concentration and downregulation of EAAT2 levels, as well as the development of mechanical allodynia. This study suggested that oxaliplatin disrupts the extracellular glutamate homeostasis in the spinal cord resulting in CINP-induced behavior, and prophylactic riluzole may decrease oxaliplatin-induced mechanical allodynia [329].

Opioid-Induced Hyperalgesia (OIH)

Prolonged and/or high doses of certain opioids such as morphine and remifentanyl may also induce symptoms of hypersensitivity, termed opioid-induced hyperalgesia (OIH). OIH mouse models develop mechanical allodynia, thermal hyperalgesia and reduced spinal GLT-1 expression in mice. Additionally, ceftriaxone reversed the downregulation of GLT-1 expression induced by repeated morphine administration and attenuated symptoms of hyperalgesia [330]. To investigate the mechanisms of OIH development, remifentanil infusions in rats were used, which produced a paradoxical increase in sensitivity to painful stimuli after prolonged opioid exposure. In this study remifentanil induced significant postoperative hyperalgesia, increased IL-1β and phospho-NR1 levels and activated the NLRP3 (NLR family pyrin domain containing 3) inflammasome by increasing toll-like receptor 4 (TLR4), purinergic P2X7 receptor (P2X7R); NLRP3, and caspase-1 expression. On the NR1 subunit of NMDARs, phosphorylation was regulated, and expression of GLT-1 was decreased in the L4-L6 spinal cord 48 h after the remifentanil infusion. This study suggests that inhibiting the NLRP3 inflammasome activation and modulating GLT-1 could be an effective for treatment of OIH [331].

3.6.3. Disease-Induced Models

Disease-induced models study neuropathic syndromes that stem from common diseases. One of the most common diseases leading to peripheral neuropathy is diabetes, which can be modeled with the pancreatic β-cell toxins streptozotocin and alloxan. In addition, pain from multiple sclerosis, end-stage cancer pain, HIV-induced neuropathy, post-herpetic neuralgia, and uremic peripheral neuropathy can also be modeled in animals to study chronic pain. Though not NP, animal models of visceral pain (colitis) were also included to further discuss EAAT regulation in chronic pain.

3.6.3.1. Diabetic Peripheral Neuropathy

Zucker Diabetic Fatty (ZDF) Rats

Zucker diabetic fatty (ZDF) rats have a missense mutation in a gene encoding the leptin receptor (fa/fa) that causes hyperglycemia, insulin resistance, hypertension and obesity, similar to type-2 diabetes [332]. Using ZDF rats, one study investigated the effects of the development of type 2 diabetes on glutamate homeostasis. These rats developed mechanical hyperalgesia and allodynia, which were associated with higher levels of basal extracellular glutamate and decreased levels of GLT-1 expression in the dorsal horn of the spinal cord. Additionally, administration of ceftriaxone prevented the development of pain behaviors that was correlated with enhanced GLT-1 expression without altering the basal glutamate levels in the spinal cord. These results suggested that impaired glutamate reuptake in the spinal cord may contribute to the development of NPs in type 2 diabetes [333].

Streptozocin (STZ)-Induced Diabetes Model

Streptozocin is an antibiotic that destroys islet β cells in the rodent pancreas to produce a model of type-1 diabetes [334]. A study on the diabetic streptozocin (STZ)-induced NP model in rats investigated the effects of clavulanic acid several days after STZ administration. On day 21, diabetic animals showed substantial mechanical allodynia, cold allodynia, and thermal hyperalgesia. Clavulanic acid reduced symptoms of allodynia and hyperalgesia, in both prophylactic and therapeutic regimens. Levels of iNOS, TNF-α, bax/bcl2 were found significantly overexpressed in spinal cord of diabetic animals, however, clavulanic acid attenuated these levels. GLT-1 expression was decreased in the spinal cord of diabetic animals but was reversed by clavulanic acid. This data suggests that clavulanic acid might be a candidate for relieving NP in diabetes mellitus, and its benefits may be attributed to the increased expression of GLT-1 [335].

3.6.3.2. Experimental Autoimmune Encephalomyelitis (EAE) Model of Multiple Sclerosis

Experimental autoimmune encephalomyelitis (EAE) in rodents is a common experimental model for the human demyelinating disease, multiple sclerosis, with some common key pathologic features of inflammation, axonal loss, and gliosis [336]. In EAE models, spinal levels of EAAT2 were decreased, which was associated with various pain behaviors. Treatment with ceftriaxone increased EAAT2 levels and alleviated hyperalgesia [337].

3.6.3.3. Visceral Pain

Visceral pain is distinctly different from the mechanism and expression of somatic pain and arises from the internal organs, particularly the gastrointestinal tract (GIT) [338]. Common GIT disorders include irritable bowel syndrome and inflammatory bowel disorders such as Crohn’s disease and ulcerative colitis. Glutamate is known to mediate visceral nociceptive neurotransmission and hypersensitivity. The role of EAAT2 was examined in an animal model of visceral pain seen in disease such as colitis. A 7 day administration of ceftriaxone mitigates visceral nociception [296], and intrathecal delivery of DHK, a pharmacological blocker of EAAT2, reversed the effect of ceftriaxone [296]. Overexpression of GLT-1 in the spinal cord with an adenovirus attenuated visceral nociception, indicating that targeting of EAAT2 could be an efficacious approach for treating visceral pain.

In conclusion, various animal models of pain were used to investigate the temporal regulation of spinal EAATs and the mechanisms involved in these changes. Some described an initial upregulation followed by downregulation of EAAT1 and EAAT2, suggesting a compensatory response, whereas other studies reported downregulation only. Some studies reported upregulation of EAAT3, a neuronal EAAT, which coincided with the downregulation of glial EAATs. The models used and the timelines investigated were not uniform, making it difficult to make conclusions on how neuronal and glial EAATs are regulated in NP models. However, collectively, these studies suggest that augmenting the expression and activity of glutamate transporters could facilitate anti-nociception in rodent NP models, perhaps when administered at specific time points. Targeting astrocytic EAATs specifically may be a promising strategy for pain therapeutic development.

3.7. Sex Differences in Neuropathic Pain

The prevalence of chronic pain and NP is higher in women, and quantitative sex differences in animal studies mirror this tendency, as female animals showing greater sensitivity to acute and chronic pain. On the other hand, sex differences in emotional-affective and cognitive responses to pain are not as well established [339]. Several animal studies that suggest a female predominance in pain hypersensitivity when compared to males. Female mice had increased mechanical allodynia and increased cold sensitivity relative to males in a model of HIV-1-associated NP [340]. In other studies, females had increased cold allodynia in a CIPN model [341], and female rats exhibited increased nociceptive sensitivity following injection of pro-inflammatory CFA into the hind paw [342]. Female mice demonstrated increased mechanical and cold hypersensitivity in the EAE model of multiple sclerosis [343]. In a CRPS model, female mice had earlier and more robust mechanical allodynia [344]. Lastly, female animals had earlier pain emergence in a model of cancer-induced bone pain [345]. These animal studies align with clinical reports of increased pain sensitivity in female subjects. However, the mechanisms behind this are not well established [346, 347].

Additionally, analgesics are not equally effective in different types of pain and between sexes [348]. Unfortunately, the studies investigating the differential, temporal expression of EAATs in NP models were mostly conducted in male rodents, so the relationship between EAAT expression, the induction and maintenance of NP, and sex differences, are not well understood. Key differences between sexes and pain development surround variances at the molecular, cellular, and systems levels of pain processing, which are likely associated with factors such as neuroinflammation, hormonal status, estrous cycles, and early life stress[339].

3.7.1. Neuroinflammation/Immunity

Both preclinical and clinical studies demonstrated that pain-related peripheral and central sensitization are influenced by neuronal communication and by elaborated crosstalk between immune cells, glia, and neurons [243, 349–351]. It is increasingly recognized that sex differences in pain could be associated with neuroimmune and glial mechanisms, which are key players in the neuroglial maladaptive plasticity that transitions acute pain to chronic pathological pain [31, 352–356].

Specific immune cells release pro-nociceptive cytokines that influences neural firing, while T lymphocytes secrete factors that support nerve repair or exacerbates nerve damage. In turn, modulating specific immune cell populations could promote nerve recovery in a sex-specific manner. This highlights a crucial need for further studies of the neuro-immune-endocrine crosstalk in the context of sexual dimorphisms in chronic pain [355, 357]. Other theories involve crucial differences in innate and adaptive immune responses between men and women [358].

A study using the CCI model revealed that female mice had allodynia for much longer than males following nerve lesion development. However, the recovery from NP was never complete, indicating that this difference could occur in the very early stages after injury. This study also revealed that male mice developed gliosis after only a few days after ligation, while females showed a late activation of glial cells that persisted for a long time [359]. In addition, in NP conditions there is a higher infiltration of T cells into nerves in females, while males had a higher infiltration of macrophages[360]. Sex differences were also found in the levels of cytokines and chemokines in blood and nerve lysates. Overall, there are important sex-associated inflammatory profiles in neuropathy that could be important for the development of differential biomarkers and sex-specific personalized medicine [361].

Previous studies suggested that spinal microglia regulate pathological pain in males [362–364]. A subsequent study examined sex-dependent glial and microglial signaling during pain in males and females. Several microglial and astroglial modulators of inflammation and NP were injected intrathecally in mice of both sexes: microglial inhibitors minocycline and ZVEID (a caspase-6 inhibitor), astroglial inhibitors L-α-aminoadipate (L-AA, an astroglial toxin), carbenoxolone (a connexin 43 inhibitor), U0126 (an ERK kinase inhibitor) and D-JNKI-1 (a c-Jun N-terminal kinase inhibitor). The results showed that spinal administration of minocycline or ZVEID, or Caspase 6 deletion, reduced formalin-induced inflammatory and nerve injury-induced NP primarily in male mice. On the other hand, intrathecal L-AA reduced NP but not inflammatory pain in both sexes, whereas intrathecal U0126 and D-JNKI-1 reduced NP in both sexes. Nerve injury resulted in upregulation of spinal GFAP and Connexin 43 in both sexes. This study revealed sex-independent astroglial signaling and confirmed male-dominant microglial signaling in the spinal cord in inflammatory and NP [365]. Toll-like receptor 9 (TLR9)-deficient males did not exhibit upregulated pronociceptive inflammatory mediators in the DRGs or increased mechanical allodynia after pain induction in the CIPN model, these effects were not observed in females [360]. This data supports the suggestions that male rodents rely on microglia-related pain processing mechanisms to a much greater extent than females. The notion that microglia-related pain processing may be a more relevant mechanism in males than females is also further supported in a study where it was observed that inhibition of the proinflammatory tumor necrosis factor receptor 1 (TNFR1) was therapeutic for CCI-induced NP in male but not female mice, as measured by NMDA receptor changes in the cerebral cortex and spinal cord [366].

Despite the rapid growth of novel findings in pain research, mechanisms of sex differences in the context of neuroimmune modulation and EAATs modulation have yet to be clearly characterized. Further understanding of factors contributing to these differences will unquestionably aid in the future development of tailored therapies for both sexes.

3.7.2. Hormonal Status, Menstrual/Estrous Cycle & Early-Life Stress

Other factors that might be associated with NP prevalence in women include hormonal status and early-life stress [367, 368]. Pain sensitivity can change throughout the menstrual cycle likely through fluctuations in estrogen levels [368]. See figure 4 for an overview of fluctuations in estrogen and progesterone in humans and rodents. In male rats, estradiol attenuates SCI -related central NP, by decreasing glutamate levels in the thalamic VPL nucleus [369]. Several studies have demonstrated that estrogen modulates the activity of sensory neurons in peripheral and central sites in female, male, or castrated animals[370]. Intravenous administration of 17β-estradiol (E2) was shown to decrease mechanical allodynia and thermal hyperalgesia in rats following SCI, it is thought that the mechanism involves inhibition of astrocyte and microglia activation [371]. Hormonal treatment with E2 resulted in complete functional recovery in the injured limb in the CCI model [372]. Similarly, subcutaneous administration of progesterone prevented mechanical and thermal allodynia in male rats following SCI rats, and decreased levels of the pro-inflammatory cytokines IL-1β, IL-6, and TNF⍺ in the spinal cord dorsal horn, supporting the notion that progesterone has a capability of suppressing reactive gliosis following CNS injury [373].

Figure 4. Fluctuations in estrogen and progesterone in humans and rodents.

Fluctuations of estrogen (i.e., 17B-estradiol; red line) and progesterone (dark blue line) across the ~ 28-day human menstrual cycle (top panel) and the ~4-day rat (middle panel) and mouse estrous cycle (bottom panel). Human menstrual cycle begins with a follicular (menstrual and preovulatory) phase, followed by ovulation, and ends with a luteal phase (spanning end of ovulation to pre-menstrual phase). In rodents, ovulation occurs every 4–5 days. Metestrus and diestrus are characterized by low increasing levels of estradiol (with slight differences between rats and mice). In the end of proestrus, estradiol is elevated, leading to a bolus of GnRH release from the hypothalamus and induction of LH and FSH surge, and ovulation occurs 12–14 h later in the estrus phase.

Created with BioRender.com

Sex differences in hyperalgesia and allodynia were examined in ligated and unligated males, and ovariectomized and intact females in the CCI model. In the absence of nerve injury, gonadally intact males had a faster response to a thermal nociceptive stimulus, compared to females. Gonadally intact and gonadectomized male and female rats developed thermal hyperalgesia within 14 days post-CCI, suggesting that the onset of the behavioral manifestations did not differ as a function of sex or hormonal status. The latency of paw withdrawal in gonadally intact males returned to baseline control values 14 days after ligation, whereas gonadally intact females, ovariectomized females and castrated males continued to elicit robust thermal hyperalgesia throughout day 35. This suggested that the mechanisms underlying chronic nociceptive processing differ as a function of sex and gonadal hormone status [346].

To examine sex differences in allodynia, a model of graded allodynia was established in rats of both sexes by altering the number of sutures placed around the sciatic nerve, and quantifying pain using a modified von Frey test with lower-graded filaments. With this approach, these investigators revealed significant sex and estrus cycle differences and concluded that females were more sensitive to mechanical allodynia throughout the diestrus and pro-estrus phases [374]. Collectively, these studies suggest sex differences in pain associated with hormonal status, nonetheless, the mechanisms involved are not clear, therefore it is critical that ongoing investigations continue to unravel such processes.

Sex differences in pain may correlate with fluctuations in EAATs expression and function, which may be mediated by estrous cycle and early life stress[368, 375]. An early study investigated sex differences and EAAT regulation by analyzing the influence of the estrous cycle on glutamate transport in rat brain synaptosomes[375]. The highest affinity for glutamate transport was observed on the proestrus and estrous phases of female rats, when compared to diestrus and in male rats. On the other hand, the maximal velocity (V_max) of glutamate transport was higher in females during diestrus compared with estrous, and lower in males compared with females in diestrus and metestrus. Exogenous application of estradiol and progesterone to synaptosomes from male rats did not alter glutamate or aspartate transport and did not alter glutamate currents in oocytes expressing EAAT1 or EAAT2. In conclusion, estrous cycle and sex differences were observed, with hormonal regulation of glutamate transporters over the four-day estrous cycle in synaptosomes from rat cortices. This regulation is unlikely due to a direct effect of estradiol or progesterone on glutamate transporters.

In an effort to correlate sex differences in EAATs expression in the spinal cord and the estrous cycle, Sprague–Dawley male and female rats were used to assess the activity of EAATs using an ex vivo aspartate radioactive uptake assay in the lumbosacral spinal cord. EAAT activity was lower in females, particularly during the estrus phase, which was the only phase that was responsive to the activator effects of riluzole on EAAT expression. The expression levels of estrogen receptors (ERα and Erβ) also showed a cycle-dependent pattern that may affect EAATs function and expression, suggesting that spinal EAAT activity in females is different from males and varies across the estrous cycle [368].

Early-life stress and the estrous cycle have been shown to influence EAATs and pain sensitivity in female Sprague-Dawley rats [367]. In model of early-life stress, animals undergo maternal separation in early life or remained non-separated. Colorectal distension (CRD) was used as a model for visceral pain, there was a decrease in glutamate uptake in the spinal cord and ACC, and increased pain behaviors, which were exacerbated in the rats that were separated from their mothers across all stages of the estrous cycle. Additionally, cortical EAAT function was decreased in separated rats during low estrogen states, however, the opposite occurred in non-separated rats. Estrous cycle and early-life stress are significant factors in visceral sensitivity, and fluctuations in EAAT function may be a factor that mediates central sensitization, likely through aberrant central circuitry involving both the ACC and the lumbosacral spinal cord.

Though these studies provided important insights about the role of sex differences, hormones, and early-life stress in NP, we still do not fully understand how these factors are linked to the regulation of spinal EAATs. Compensatory mechanisms are likely involved, and more studies are needed to unravel sex-specific targets for future analgesic drug discovery.

4. Emerging therapies through the modulation of glutamate transporters

While numerous mechanisms are explored for NP therapy, the modulation of the glutamatergic system to decrease pain transmission continues to hold promise. Glutamate receptor antagonism provides anti-nociception; however, it may be associated with severe side effects including sedation, hallucinations and cardiac instability limiting its use in outpatient settings [107, 108, 376]. A potential new therapeutic approach is the modulation of glutamate transporters, such as EAAT2, located on astroglial cells surrounding glutamatergic neurons to restore homeostasis by lowering extracellular glutamate concentrations [185, 186, 377]. Downregulation or inhibition of EAATs has been shown to increase pain [185, 203], whereas increased EAAT2 expression can mitigate pain. Furthermore, positive allosteric modulation (PAM) of glutamate transporters can increase the efficiency of remaining EAATs in various neurological diseases offering neuroprotection [378–380] and is a promising future target for chronic pain.

4.1. Drugs that increase EAAT expression

There are several drug classes that have been found to have analgesic properties in animal models of pain through the modulation of glutamate transporters. Augmentation of the expression of glutamate transporters could facilitate anti-nociception in rodent NP models, presumably by increasing the removal of excessive glutamate. These drugs include tetracycline antibiotics (e.g., minocycline), β-lactam antibiotics (e.g., ceftriaxone), β-lactamase inhibitors (e.g., clavulanic acid), anticonvulsants (VPA and riluzole), and tricyclic antidepressants (amitriptyline).

4.1.4. Antibiotics

Ceftriaxone

Ceftriaxone was the first compound identified to increase the expression of GLT-1 [290], prompting further research of this drug as a potential novel therapy to modulate GLT-1 expression. As a result, numerous studies have been conducted to examine the effect of this compound in many models of neurological disorders, including NP. Early studies demonstrated that ceftriaxone upregulated GLT-1 expression and glutamate uptake in the spinal dorsal horn and had anti-nociceptive effects in the CCI model [381]. Another group confirmed the effect of ceftriaxone on reversing tactile allodynia induced by CCI and extended the studies to include the EAE model of multiple sclerosis, in which ceftriaxone reversed tactile allodynia and halted the progression of motor weakness and paralysis. They also observed that EAE and CCI each significantly reduced the expression GLT-1 and induced astrocyte activation in the lumbar spinal cord, which were reversed by ceftriaxone, suggesting that this drug suppresses glial activation and alleviates NP [382]. In a model of repeated morphine administration in mice, daily intrathecal treatment with ceftriaxone attenuated the development of OIH and allodynia, prevented astrocyte activation, and reversed the downregulation of GLT-1 expression induced by morphine [330]. Additionally, in the SNI model, ceftriaxone treatment increased GLT-1 expression in the dorsal horn, resulting in increased glutamate uptake and lower extracellular glutamate, which attenuated NP-like behavior [306]. Further studies determined that ceftriaxone had an effected in an inflammatory model with intra-plantar injections of formalin and in the SNL model. Ceftriaxone interfered in the transmission of noxious signaling and had an effect upon acute thermal and mechanical pain thresholds [383]. Using the ZDF diabetes model, intrathecal administrations of ceftriaxone prevented the development of pain behaviors in a dose dependent manner and was correlated with enhanced GLT-1 expression without altering the basal glutamate levels in the spinal cord, suggesting that impaired glutamate reuptake in the spinal cord could contribute to the development of NP in type 2 diabetes [333].

Collectively, these studies validated a potential anti-nociceptive effect of ceftriaxone in several NP models. Two additional studies investigated different questions using ceftriaxone as a tool: one focused on understanding region-specific modulations of GLT-1 levels in NP [384] and the other on unraveling mechanisms involved in EAAT2 upregulation [385]. Changes in glutamatergic transmission in NP states have been reported in the spinal cord and several supraspinal regions, such as the periaqueductal gray (PAG), and GLT-1 is reduced in the spinal cord but increased in the PAG after nerve injury, suggesting different roles of GLT-1 in NP in different brain regions. The first study generated PAG-specific and spinal cord-specific GLT-1 knockout mice that were subjected to the pSNL model. NP behavior was enhanced in spinal cord-specific GLT-1 knockout mice but alleviated in PAG-specific GLT-1 knockout mice, suggesting an enhancement of glutamatergic neurotransmission from primary sensory neurons to the postsynaptic dorsal horn. Downregulation of GLT-1 in the spinal cord results in insufficient descending pain inhibition caused by GLT-1 upregulation in the PAG resulting in NP. Additionally, ceftriaxone upregulated GLT-1 expression in the spinal cord, but not in the PAG, of control mice and, furthermore, attenuated tactile hypersensitivity in pSNL WT mice, but not in pSNL in the spinal cord-specific GLT-1 knockout mice. These results indicate that the anti-NP effect of ceftriaxone is mediated by the upregulation of GLT-1 expression in the spinal cord. Furthermore, selective upregulation of spinal GLT-1 and/or downregulation of GLT-1 in the PAG could be a potential novel strategy for the treatment of NP. The second study investigated whether the mechanism of ceftriaxone was involved in astrocytic endothelin-1 (ET-1). It is known that targeted overexpression of astrocytic ET-1 attenuates NP by upregulating spinal EAAT2, and mice overexpressing ET-1 or WT underwent SNL with or without ceftriaxone treatment for 4 days. Transgenic mice had increased levels of EAAT2 and reduced SNL-induced NP, and EAAT2 induction by ceftriaxone reduced SNL-induced NP in both mice, suggesting that overexpression of ET-1 suppressed NP by upregulating spinal EAAT2 expression via ET receptors [385].

Although these studies suggest anti-nociception via upregulation of EAATs with ceftriaxone in several NP models, its clinical effectiveness has yet to be elucidated. There are few clinical trials looking at ceftriaxone as a possible analgesic therapy. In one double blind RCT, a single dose of ceftriaxone, cefazolin (negative control), or saline was administered to 45 patients undergoing decompressive median or ulnar nerve surgery [386]. Patients on chronic pain therapy were excluded from the trial, and those on NSAIDs discontinued therapy several days prior to their procedure. The procedure was performed under local anesthesia with 1% mepivacaine, and mechanical pain thresholds using von Frey filaments were measured 10 minutes prior to surgery and 4–6 hours postoperatively. No other analgesics drugs were administered before the postoperative pain analysis. The results of this study demonstrated a 10-fold increase in pain thresholds (significant decrease in pain) in patients who received ceftriaxone compared to those who received cefazolin or saline [386].

Clavulanic acid

Concerns over the antibiotic activity of ceftriaxone led to the study of clavulanic acid, a β-lactamase inhibitor that is a structural analog of ceftriaxone and also enhances GLT-1 expression, but lacks antibiotic activity and has better brain penetrability and oral activity [387]. In the CCI model, clavulanic acid and ceftriaxone demonstrated anti-nociception [388], increased GLT-1b expression within the rat spinal cord [389] and reversed the down regulation of GLT-1 at day 14 following surgery [390]. Ceftriaxone and clavulanic acid were also shown to have anti-allodynia properties, measured by mechanical and thermal responses, and anti-inflammatory effects, measured by analysis of TNF-α and IL-10 serum concentrations, in rats using the carrageenan inflammatory model [391]. A recent study demonstrated that clavulanic acid reduced symptoms of allodynia and hyperalgesia in the diabetic STZ-induced NP model in rats, as well as reversed GLT-1 decreased expression in the diabetic animals. Clavulanic acid also decreased levels of iNOS, TNF-α, bax/bcl2 in spinal cord of diabetic animals, suggesting that this compound increased expression of GLT-1, and inhibited nitrosative stress (a type of oxygen metabolism disorder that induces cell death [392]), inflammation, and apoptotic mediators. This study suggests that clavulanic acid might be a candidate for relieving NP in diabetes mellitus, and its benefits might be attributed to the increased expression of GLT-1 [335]. Two additional studies also showed that when ceftriaxone or clavulanic acid was co-administered with opioids, such as morphine, or cannabinoids, a decrease in drug tolerance was observed [393, 394], suggesting a role of GLT-1 in morphine dependence and tolerance, which might be important for future drug development efforts.

A clinical trial randomized patients with chronic low back pain to a 100-day course of amoxicillin-clavulanic acid or placebo. Patients reported greater pain relief in the treatment group at 6–8 weeks after starting therapy that lasted 6–12 months after the end of the treatment period [395]. However, when this study was repeated by a different group, the results could not be replicated, and the investigators concluded that this therapy lacked efficacy in this patient population [396].

Minocycline

Few studies have investigated minocycline, a tetracycline antibiotic experimentally used to study NP therapeutics. A study in a paclitaxel-induced CINP model reported a rapid and persistent activation of spinal astrocytes, and a significant downregulation of GLAST and GLT-1 in spinal dorsal horn, which was reversed by systemic treatment with minocycline [325]. Minocycline was also shown to ameliorate both the downregulation of glial EAAT expression, and the activation of astrocytes induced by pSNL in the spinal dorsal horn [397]. Intriguingly, minocycline alone did not affect the CCI induced-thermal hyperalgesia in neuropathic animals, but in combination with ceftriaxone resulted in enhanced latency time to noxious thermal stimulus[398], suggesting that modulation of microglia activity could have a supportive role in the improvement of CCI-induced thermal hyperalgesia. Conversely, another study reported minocycline had an anti-nociceptive effect in the CCI model, but only when administered prophylactically before the surgery [399]. A systematic review of human studies showed efficacy of minocycline in controlling pain from diabetic and leprotic neuropathies[400].

4.1.4. Other (LDN-212320)

A screen approach was performed to identify compounds that increase EAAT2 expression, which resulted in the identification of compound LDN-212320 (3-[[(2-Methylphenyl) methyl] thio] −6-(2-pyridinyl) -pyridazine) [401]. The mechanism of this compound is to increase the translation of EAAT2, which differs from ceftriaxone and clavulanic acid, compounds that increase EAAT2 transcription. LDN-212320 and several analogs were examined in a rat model of oxaliplatin-induced CINP. After repeated administration of these compounds, different noxious and non-noxious mechanical and thermal stimuli were performed. One of the analogs, 4f, demonstrated the best profile that fully counteracted the oxaliplatin-induced neuropathy on day 14, suggesting that translational activation of EAAT2 could be considered for further preclinical pain studies [402]. Another study centered on investigating LDN-212320 using the formalin-induced nociceptive pain model and hippocampal-dependent behavioral tasks in mice. This compound significantly attenuated nociceptive behavior, which was reversed by systemic administration of DHK, an EAAT2 antagonist. In addition, LDN-212320 increased GLT-1 expression in the hippocampus and ACC, areas involved in pain processing and modulation. The compound decreased formalin-induced ERK phosphorylation, a marker of nociception, providing evidence that EAAT2 activation could be a potential therapeutic for the treatment of pain [321]. Further studies with translational activators of EAAT2 are needed, and no clinical trials have been performed with LDN-212320 or analogs to date.

4.1.3. Antidepressants

Amitriptyline is a tricyclic antidepressant shown to increase EAAT2 expression in preclinical studies. In the SNI model, amitriptyline induced NF-KB (nuclear transcription factor-kappa B)-dependent upregulation of EAAT2 [403], reversed the downregulation of EAAT1 and EAAT2 and attenuated mechanical allodynia [304]. However, a systematic review and meta-analysis including 17 studies suggested amitriptyline will not be efficacious for NP treatment [83].

4.1.4. Anticonvulsants

4.1.4.1. Valproic acid (VPA)

The precise mechanism of analgesia from VPA is unclear, but it may result from blockage of voltage-dependent sodium channels, increased EAAT expression [98], and inhibition of HDAC I and II [404]. In the SNL model there is a significant reduction of mechanical withdrawal thresholds and expression of GLT-1 and GLAST in the spinal dorsal horn. Administration of VPA reduced hypersensitivity and restored down-regulated expression of GLT-1 and GLAST in the spinal dorsal horn, which was blocked by DHK. This suggests that valproate restores the downregulated expression of glutamate transporters in the spinal cord to reduce glutamate signaling and hypersensitivity after nerve injury. The study also combined VPA with riluzole and concluded that enhanced analgesia relies on the spinal glutamate transporters [100]. Another study from this group supported that VPA prevents downregulation of EAAT1 and EAAT2 in primary cultured astrocytes and in rats after SNL, which also showed reduced hypersensitivity after compound administration. These effects were blocked by intrathecal administration of GLT-1-selective small interfering RNA (siRNA), suggesting a direct connection. Microanalysis measurement of extracellular glutamate concentration in the spinal cord revealed increased levels in animals with SNL or after GLT-1 selective siRNA treatment that was prevented by VPA [102].

These studies suggest that VPA reduces the development of chronic pain after nerve injury partially by preventing the downregulation of GLT-1 and maintaining glutamate homeostasis in the spinal cord. Epigenetic changes, especially by HDACs, are also important mechanisms in neuroinflammation and neuroglia plasticity in the spinal cord after inflammation and nerve injury.

Though VPA can be prescribed for NP and fibromyalgia, it’s side effects and limited efficacy preclude regular use in this population. Compared to placebo, study participants on VPA for diabetic neuropathy or post-herpetic neuralgia experienced more nausea, drowsiness, and abnormal liver function tests [98].

4.1.4.2. Riluzole

Riluzole, an FDA-approved drug for the treatment of ALS, has been demonstrated to attenuate neural excitotoxicity by blocking glutamate receptors overactivation through prevention of glutamate release from presynaptic terminals by inhibiting voltage-gated Na⁺ and Ca²⁺ channels [101, 105, 405], and by elevating GLT-1 expression and activity on striatal astrocytes [406]. Riluzole has also been investigated for antinociceptive effects. In SCI, a single systemic administration of riluzole reversed cutaneous hypersensitivity in rats and, furthermore, this antinociception appeared to be mediated peripherally [407]. Another study in the rat CCI model revealed that administration of riluzole decreased mechanical allodynia and thermal hyperalgesia and suggested the mechanism involved downregulation of purinergic ionotropic receptor P2X7R expression and inhibition of microglial activation in the spinal cord dorsal horn [408]. In the oxaliplatin-induced peripheral neuropathy model, riluzole suppressed the increase of glutamate concentration, reverted the decreased GLT-1 expression in the spinal cord and the development of mechanical allodynia [409]. In the SNI model, riluzole had beneficial effects on NP behaviors through the modulation of calcium-activated potassium channels in the amygdala, a brain region involved in pain, and emotional-affective states and disorders [410]. It remains to be determined whether riluzole can be used clinically for NP, however, no clinical trials for this indication are currently being performed.

4.1.5. Non-Pharmacologic Therapies

Electroacupuncture has antinociceptive effects via inhibition of GLAST and GLT-1 downregulation in the CFA rat model. Expression of GFAP was significantly increased in CFA-injected rats, which was inhibited by electroacupuncture stimulation. On the other hand, downregulation of GLAST and GLT-1 was observed in CFA-injected rats on days 2 and 3, and electroacupuncture stimulation recovered their expression, providing further evidence for the involvement of astrocytic glutamate transporters in response to activation of spinal astrocytes [411]. In the SNI model, GLAST and GLT-1 levels were decreased at days 8, 14 and 20 post surgery. Similar to the CFA model, there was a significant upregulation in the transporter levels and attenuated pain behaviors after electroacupuncture treatment. Intrathecal injection of glutamate-transport inhibitor, PDC, reduced this analgesic effect, confirming that electroacupuncture can increase the astrocytic EAAT levels in NP rats [412]. The same group also explored the role of spinal EAATs in electroacupuncture tolerance, which was induced by treating rats with electroacupuncture once a day for 8 days, followed by intrathecal administerion of riluzole to increase EAAT expression. L4–5 spinal cords were collected at days 0, 2, 4, 6 and 8 after repeated electroacupuncture. They observed higher expression levels of the spinal EAATs at days 2 and 4, which gradually decreased as the times of electroacupuncture increased. At day 8, no difference was observed in the spinal EAATs between the sham and electroacupuncture treatment. Intrathecal administration of riluzole dose-dependently attenuated the decreased electroacupuncture analgesia, strongly suggesting involvement of spinal EAATs in the electroacupuncture effect [413].

4.2. Novel compounds that increase transport efficiency through allosteric modulation

Emerging compounds that allosterically modulate EAAT2 have demonstrated neuroprotective properties in basic studies. PAMs of EAAT2 [379, 380] increase the uptake of glutamate from the extracellular space, subsequently decreasing excitotoxicity, and are potential therapies for a variety of neuropathologic conditions including stroke, brain trauma, epilepsy, NP, among others. In in vitro models of glutamate-mediated excitotoxicity, EAAT2 PAMs demonstrated neuroprotection in primary rodent neuron-glia cultures [380]. Because EAAT2 downregulation is repeatedly demonstrated in the pathology of NP, increasing the activity with a PAM of the remaining transporters could be a promising new avenue for future drug discovery. Furthermore, selective EAAT2 PAM differs considerably from NMDAR antagonists, since its mechanism is to restore glutamate homeostasis by increasing uptake of excessive glutamate, not block glutamatergic transmission, and is expected to be safe and devoid of the dissociative effects observed with NMDAR antagonists. We expect that these compounds will differ considerably from transcriptional or translational upregulators of EAAT2 expression, such as ceftriaxone, that has demonstrated modest benefit in preclinical pain assays [330, 381, 389]. EAAT2 PAMs also vary from pharmaco-epigenetic strategies to activate EAAT2 expression, such as VPA, that would differ across individuals and raise concerns over safety, efficacy, and adverse effects[98]. Because EAAT2 PAMs work directly on the transporter, they rapidly increase glutamate uptake efficiency, have a quicker onset compared to drugs that modulate expression, and are highly selective offering a more robust and safe clinical profile compared to epigenetic-modifying expression enhancers that are often associated with adverse side effects.

Currently, our group studies the potential antinociceptive properties of EAAT2 PAMs in animal models of NP. Our scientific premise is that EAAT2 PAMs will enhance glutamate uptake and correct aberrant glutamate signaling in the spinal cord to reduce pain signaling from the periphery to the spinal cord [414]. Our preliminary data (unpublished) of our lead EAAT2 PAM demonstrates antinociceptive activity in both the SNI and SNL models. We are currently completing studies to clarify the temporal expression levels of astrocytic EAAT1 and EAAT2 and neuronal EAAT3 transporters in the spinal cord and brains of both male and female rodents subjected to SNI. We expect that PAM administration will have an effective antinociceptive effect at specific doses, though, we anticipate that the best timing and dose-dependence will vary between sexes. We also expect that expression of EAATs will fluctuate in a time-dependent manner due to differences in compensatory mechanisms. These studies will provide insights on the therapeutic potential of directly targeting this transporter through allosteric modulation.

In addition to being a novel target for treating pain, glutamate transporters are also implicated in addictive behaviors [198, 415]. A drug that reduces chronic pain without the risk of addiction would truly bridge the practice gap. Our studies aim to investigate whether these EAAT2 PAM compounds could modulate opioid-seeking behavior and relapse in naïve and animals with NP, in males and females. These studies will provide the in vivo proof-of-concept that the strategy of modulating glutamate levels can be developed for NP therapeutics and for opioid dependence, two independent diseases that may develop as co-morbidities in patients who failed first- and second-line NP therapies and resorted to opioid use.

It is important to assess the safety and side effects of modulating EAAT2 activity globally by performing abuse liability tests, neurological measures, off-target assessments, toxicity, and toxicokinetic studies. Future studies will also focus on improving the drug properties including the pharmacokinetics of these compounds. Ultimately, this work will guide the drug-development process of EAAT2 PAMs as a therapy for NP.

5. Conclusions

Chronic pain affects 1 in 5 people and can influence a person’s physical health and their psychosocial and mental well-being. NP is of special interest because it is difficult to treat and remains refractory to many opioid and non-opioid therapies currently on the market, including antidepressants, anticonvulsants, NMDAR antagonists, morphine, and topical and localized therapies. Furthermore, for many patients that ultimately resort to opioid therapy, there is the pervasive concern that treatment can possibly lead to adverse effects and addiction. Nociception occurs through a complex interaction of cellular and molecular components involving neurons, glia, glutamate receptors and transporters that use L-glutamate, the major transmitter released by sensory afferents in the nervous system. This system is dysregulated in the establishment of NP, which is validated by several preclinical studies suggesting that dysregulation of EAATs correlates with increased nociceptive behavior, and increased expression of these transporters is associated with decreased nociception. Nevertheless, precisely how the temporal regulation of glutamate transporters occurs in the development and establishment of NP is not fully elucidated. Animal NP models display diverse results: some indicate early upregulation followed by later downregulation of glial transporters while others suggest only downregulation occurs accompanied by changes in neuronal transporters, suggesting compensatory mechanisms. Further studies are needed to understand how the glutamate transporters are regulated in the acute and chronic phases of pain. Additionally, chronic pain and NP are prevalent in females, and this pervasiveness may involve neuroinflammation/immunity factors as well hormonal status and early-life stress components, with some studies suggesting a correlation with dysregulated EAAT levels. Studies are needed to fully characterize mechanisms of sex differences in the neuroimmune modulation and EAATs sex-specific pathways at all levels of the neuraxis. Lastly, therapies involving the modulation of glutamate transporters hold promise for future development of NP therapeutics. Drugs increasing EAAT expression have demonstrated anti-nociceptive activity in numerous animal studies. Unfortunately, human trials on these medications are limited, or have demonstrated low efficacy and adverse effects. PAMs of EAAT2, the predominant glutamate transporter in the CNS, has demonstrated neuroprotection in excitotoxicity models and may offer analgesic benefit in NP models. Altogether, there remains a great need to develop safe and efficacious drug therapies for this debilitating condition.

Abbreviations

ACC

anterior cingulate cortex

ALS

amyotrophic lateral sclerosis)

AMPA

α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid

AMPAR

α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor

AP-V

2R)-amino-5-phosphonovaleric acid

CBD

Cannabidiol oil

CBM

Cannabis & Cannabis-based Medicines

CB₁R, CB₂R

cannabinoid receptors 1 and 2)

CCI

chronic constriction injury

CDC

Centers for Disease control

CFA

Complete Freund’s Adjuvant

CIPN

chemotherapy-induced peripheral neuropathy

CNS

Central Nervous System

CPIP

chronic post-ischemia pain

CRPS

chronic regional pain syndrome

DHK

dihydrokainate

D-JNKI-1

c-Jun N-terminal kinase inhibitor

DL-TBOA

DL-threo-β-benzyloxyaspartate

DRG

dorsal root ganglia

EAAC1

excitatory amino acid carrier 1

EAATs

excitatory amino acid transporters

EAAT1–3

human excitatory amino acid transporter subtypes 1–3

EAE

experimental autoimmune encephalomyelitis

eCS

endocannabinoid system

E2

17β-estradiol

GFAP

glial fibrillary acidic protein

GAD

glutamate decarboxylase

GIT

gastrointestinal tract

GLAST

glutamate and aspartate transporter

GLT-1

rat glutamate transporter 1

GlyT1

glycine transporter 1

GluRs

glutamate receptors

GSK3β

glycogen synthase kinase 3beta)

HDAC

histone deacetylase

HPA

hypothalamus-pituitary-adrenal

HIV

human immunodeficiency virus

IASP

International Association for the Study of Pain

iGluR

ionotropic glutamate receptor

ICD

International Classification of Diseases

IL-1β

Interleukin 1 beta

IV

intravenous

KBBP

kappa B-motif binding phosphoprotein

L-AA

L-α-aminoadipate

LDN/OSU-0212320

3-[[(2-Methylphenyl)methyl]thio]-6-(2-pyridinyl)-pyridazine

lipopolysaccharide

LPS

mEPSCs

miniature excitatory postsynaptic currents

miRNA-107

micro RNA-107

LC

Locus coeruleus

mGluR

metabotropic glutamate receptor

NP

neuropathic pain

NLRP3

NLR family pyrin domain containing 3

NMDA

N-methyl-D-aspartic acid

NMDAR

N-methyl-D-aspartic acid receptor

NRC

Nerve Root Compression

OIH

opioid-induced hyperalgesia

OPCs

Oligodendrocyte progenitor cells

PAG

periaqueductal gray

P2X7R

purinergic P2X7 receptor

pSNL

Partial Sciatic Nerve Ligation

PAM

positive allosteric modulation, or modulator

PDC

pyrrolidine-2,4-dicarboxylic acid

RCT

randomized controlled trial

RVM

rostral ventromedial medulla

SAHA

suberoylanilide hydroxamic acid

SCG

satellite glial cells

SNRI

serotonin-norepinephrine reuptake inhibitor

SCI

Spinal cord injury

SNI

spared nerve injury

SNL

spinal nerve ligation

sPLA2

spinal secretory phospholipase A2

STT

spinothalamic tract

STZ

Streptozotocin

TCA

tricyclic antidepressants

TF-TBOA

(3S)-3-[[3-[[4-(Trifluoromethyl)benzoyl]amino]phenyl]methoxy]-L-aspartic acid, TFB- TBOA

TLR4

Toll-like receptor 4

TLR9

Toll-like receptor 9

TNF α

Tumor necrosis factor α

U0126

1,4-Diamino-2,3-dicyano-1,4-bis[2-aminophenylthio]butadiene

vGAT

vesicular GABA transporter

vGluT

vesicular glutamate transporter

VPA

valproic acid

xCT

glutamate-cystine exchangers

YY1

Yin Yang 1