Introduction

Pain, as defined by the International Association for the Study of Pain, is an unpleasant sensory and emotional experience associated with, or resembling that associated with actual or potential tissue damage. Under physiological conditions, pain plays a protective role to warn the organism to evade noxious stimuli (such as heat, chemical irritants, and cold) and avoid them in future. However, under injury or disease conditions, pain can persist for months to years, and this type of pain is called pathological or chronic pain. Chronic pain is characterized by spontaneous pain, allodynia (pain evoked by a normally innocuous stimulus), and hyperalgesia (enhanced pain evoked by a noxious stimulus). Changes in neuronal plasticity are the major mechanisms of chronic pain [1]. Thus, several neuron-targeting drugs such as NMDA receptor antagonists, opioids (such as morphine, oxycodone, and codeine), and Na+ channel blockers (such as lidocaine, oxcarbazepine, and carbamazepine) are used for the treatment of chronic pain. Although these drugs have therapeutic effects, they also have different degrees of side-effects [2]. Therefore, the development of new types of analgesic with few adverse reactions and better targeting are urgently needed. In the last two decades, non-neuronal cells, especially glial cells, have attracted increasing attention. Targeting the function of glial cells is likely to be a new direction for chronic pain treatment.

In the central nervous system (CNS), more than half of the cells are glia (including astrocytes, microglia, and oligodendrocytes), ~20%–40% of which are astrocytes [3, 4]. Astrocytes not only provide structural and nutritional support for neurons, they also play important roles in many neural processes [5]. Under normal conditions, astrocytes are mostly in a resting state; however, when tissue injury or disease occurs, they transform into a reactive state and contribute to the development of neurological disorders. One of the important features of astrocytes, different from other glial cells, is that they directly communicate with each other by forming gap-junction protein complexes, which allow adjoining cells to freely exchange ions and small cytosolic components [6]. In addition, when CNS neurons are activated, astrocytes regulate blood flow through their extensive contact with cerebral blood vessels [7]. Astrocytes also widely connect with neuronal synapses: a single cortical astrocyte can contact 4–6 neuronal somata, almost 140,000 synapses, and 300–600 neuronal dendrites [3, 4]. Close contact with neurons and synapses makes it possible for astrocytes to support neurons and regulate the physiological/pathological environment during synaptic transmission. These features show that astrocytes play important roles in signal transmission and processing.

In this review, we provide an overview of the roles of astrocytes in the pathogenesis of chronic pain and the interactions between astrocytes and microglia/neurons. We discuss recent neurobiological mechanisms and possible downstream molecular pathways of the astrocytic control of chronic pain. Finally, we discuss how they can be targeted as an alternative strategy for the treatment of chronic pain.

Spinal Astrocytes in Chronic Pain

Classically, chronic pain is classified into two main categories: nociceptive and neuropathic. Nociceptive pain is associated with an ongoing input from real or threatened tissue injury, such as arthritis, trauma, and visceral inflammation. Neuropathic pain is caused by a direct consequence of a lesion or disease affecting the somatosensory system, such as nerve or nerve root compression, toxins, and ischemia. In 2016, the term nociplastic pain was proposed to describe pain that arises from the abnormal processing of pain signals without any clear evidence of tissue damage or discrete pathology involving the somatosensory system, such as fibromyalgia, irritable bowel syndrome, and temporomandibular disorder [8]. In the past two decades, the role of astrocytes in nociceptive pain and neuropathic pain has been widely studied. Under various disease conditions, astrocytes are activated and change to a reactive state characterized by morphological, molecular, and functional changes. Reactive astrocytes are identified by glial fibrillary acidic protein (GFAP) upregulation and hypertrophy after nerve injuries such as spinal nerve ligation (SNL) [9,10,11,12], chronic constriction injury (CCI) [13,14,15], and spinal cord injury (SCI) [16,17,18]. In addition, this process has also been reported in other pain models like tissue inflammation [19, 20], chemotherapy-induced pain [21, 22], arthritic pain [23, 24], and chronic post-cast pain [25, 26].

Reactive astrocytes can display various states. Some reactive states happen within minutes, such as a change in the phosphorylation of signaling molecules or an increase in intracellular Ca2+. Others occur after hours or days (e.g., astrocyte hypertrophy or translational regulation). When a peripheral nerve is injured, astrocyte hypertrophy occurs 3 days later and lasts for several months [27]. Different from astrocytes, microglia immediately respond to a stimulus and proliferate [28], and this process grows to maximal levels in the first week following nerve injury. Raghavendra et al. reported that in a neuropathic pain model, when minocycline (a microglia activation inhibitor) administration is started at the time of nerve transection (pre-emptive treatment), it reduces allodynia and hyperalgesia, which is associated with its ability to suppress microgliosis. However, administration on day 5 after surgery (treatment of existing hypersensitivity) fails to attenuate the behavioral hyperalgesia and allodynia, although it inhibits microglial activation [29]. Meanwhile, astrocyte inhibitors work in both the early and late phases of neuropathic pain [53]. IL-1 receptor-knockout mice show decreased nociceptive responses after SNL [54]. In addition, optogenetic activation of astrocytes causes an increase of IL-1β and TNF-α secretion [36]. In human studies, TNF-α and IL-1β are also increased in the spinal astrocytes of patients with HIV-associated chronic pain [55]. These results suggest the important role of TNF-α and IL-1β in regulating neuropathic pain.

In addition to cytokines, several chemokines such as CCL1, CCL2, CCL3, CCL4, CCL7, CXCL10, CXCL12, CXCL13, and CX3CL1 and their receptors contribute to the pathogenesis of neuropathic pain [56]. Here, we focus on the chemokines or chemokine receptors associated with astrocytes. CCL1 was initially identified in T cells and stimulates the migration of human monocytes through binding to its receptor CCR8. In the SNL model, CCL1 is mainly produced in the DRG and transported to the spinal cord [57]. Meanwhile, CCR8 is increased in astrocytes of the ipsilateral superficial dorsal horn. Inhibition of CCL1 by intrathecal injection of a neutralizing antibody reduces nerve ligation-induced tactile allodynia [57]. Also, oral administration of RAP-103, a peptide inhibitor of CCR8, fully prevents mechanical allodynia and inhibits the development of thermal hyperalgesia after SNL, suggesting the involvement of CCR8 in the initiation and maintenance of nerve injury-induced neuropathic pain [58, 59].

CCL2 is highly expressed by spinal astrocytes and is upregulated in the SNL model [46]. Meanwhile, CCL2 is produced in cultured astrocytes after stimulation with lipopolysaccharide (LPS), TNF-α, or IL-1β [60]. CCR2 is the major receptor of CCL2 and is expressed in primary afferents and neurons in the spinal cord [61]. Mice overexpressing CCL2 in astrocytes display enhanced nociceptive responses in the CFA (complete Freund’s adjuvant) model [62]. Our previous study demonstrated that CCL2 induces rapid phosphorylation of ERK (extracellular signal-activated kinase) in spinal cord neurons. In addition, when lamina II neurons in the spinal cord slice are recorded, the application of CCL2 immediately enhances NMDA- and AMPA-induced inward currents and causes an increase in the frequency and amplitude of sEPSCs [46]. CCL2 also modulates inhibitory synaptic transmission since it inhibits GABA-induced currents in spinal neurons [63].

CXCL10 belongs to the CXC chemokine family and is also known as keratinocyte-derived chemokine-10. CXCL10 is the major ligand of CXCR3 and is increased in neurons and astrocytes of the spinal cord after SNL or spinal cord ischemia reperfusion [64, 65]. Inhibition of CXCL10 by spinal injection of shRNA lentivirus attenuates SNL-induced mechanical allodynia and heat hyperalgesia [64, 66]. CXCL9 and CXCL11 belong to the same subfamily as CXCL10 [67]. However, the roles of these chemokines in pain hypersensitivity are different from CXCL10. Intrathecal injection of CXCL9 or CXCL11 does not induce hyperalgesia or allodynia behaviors, and their inhibition does not inhibit neuropathic pain either [68], suggesting different roles of these chemokines in pain regulation. Other chemokines and their receptors such as CX3CL1/CX3CR1, CXCL1/CXCR2, and CXCL12/CXCR4 are also involved in chronic pain and have been introduced in our previous and others’ reviews [56, 69, 70], so they are not discussed in detail here.

Channel Proteins

Different types of cationic or anionic channels are located on astrocytic membranes to regulate ions for the resting membrane potential or conductance and intracellular signaling. The ion channels on astrocytes are also involved in regulating the release of various gliotransmitters associated with several physiological processes. Here, we introduce some typical water and ion channels that are expressed on astrocytes involved in chronic pain regulation.

Aquaporin-4 (AQP4) is a major water channel expressed in the central nervous system, primarily in astrocytes. The role of AQP4 has been widely studied in a range of pathological conditions [71]. In the spinal cord, AQP4 exhibits a graded decline in distribution from the dorsal to the ventral horn, with abundant expression in laminae I and II [72]. The function of AQP4 is to regulate water influx or efflux driven by osmotic pressure to maintain water homeostasis. AQP4 is increased in spinal cord astrocytes after SCI and nerve injury in humans [73]. In addition, AQP4-knock-out mice show reduced pain sensitivity and dorsal horn sensitivity to noxious stimulation [74].

Sulfonylurea receptor 1 (SUR1), encoded by the Abcc8 gene, is a regulatory subunit that co-assembles with the inward rectifier K+-selective channel to form the KATP channel [75]. SUR1 also co-assembles with the non-selective cation channel, transient receptor potential melastatin 4 (TRPM4) to form the SUR1-TRPM4 complex. SUR1-TRPM4 is upregulated in dorsal horn astrocytes, and global or astrocytes-targeted deletion of SUR1-TRPM4 relieves mechanical allodynia and thermal hyperalgesia in a sciatic nerve cuffing mouse model [76]. Meanwhile, chronic administration of glibenclamide (an SUR1 antagonist) to mice with neuropathic pain causes a reduction of pain behaviors and the expression of IL-6, CCL2, and CXCL1 in astrocytes. Thus, glibenclamide may be an astrocyte-targeted candidate drug for the treatment of some kinds of neuropathic pain.

P2X3 is a non-selective ligand-gated ion channel that belongs to the purinergic receptor family [77]. P2X3 is activated by adenosine triphosphate (ATP) and is selectively permeable to Na+, K+, and Ca2+, especially Ca2+, which plays an important role in the generation and transmission of nociceptive information [78]. Nerve injury causes the release of a large amount of ATP, which activates the P2X3 receptor in the presynaptic membrane and causes Ca2+ influx, resulting in phosphorylation of PKA and PKC and the release of glutamate. This process further activates the NMDA receptors on neurons and causes EPSC generation and central sensitization [79]. It is well known that activation of P2X3 in the DRG causes abnormal nerve discharge, strengthens the transmission of sensory information, and induces visceral hyperalgesia [80, 81]. P2X3 is also expressed on astrocytes in the spinal cord and is increased in a rat model of neuropathic pain [82]. Inhibition of P2X3 in the spinal cord reduces hypersensitivity after nerve injury. In addition, administration of MPEP (2-methyl-6-(phenylethynyl)pyridine; an mGluR5 antagonist) reduces the mechanical allodynia and abolishes the increase in the density of P2X3 in astrocytes induced by nerve injury [82].

Enzymes

Metalloproteases (MMPs) have been suggested to act in the cleavage of extracellular matrix proteins, cytokines, and chemokines to control the inflammation and tissue remodeling associated with various neurodegenerative diseases [83]. MMP-2 and MMP-9 are members of the MMP family involved in IL-1β cleavage [83, 84]. Spinal astrocytes continuously secrete MMP2 after SNL. Downregulation of MMP-2 through intrathecal injection of MMP-2 siRNA reduces mechanical allodynia and the level of spinal GFAP and phosphorylated c-Jun N-terminal kinase 1/2 (JNK1/2), an astrocyte-expressing kinase, in a neuropathic pain model. Local inhibition of MMP-9 inhibits the early phase of neuropathic pain, whereas inhibition of MMP-2 suppresses the late phase of neuropathic pain [83].

Transforming growth factor-β-activated kinase 1 (TAK-1), also known as MAPK kinase kinase 7, is an enzyme regulating innate immunity and pro-inflammatory signaling [85]. TAK-1 mediates the activation of the nuclear factor-kB (NF-kB), JNK, and p38 pathways. Soto-Diaz et al. found that both the production of chemokines and neutrophil migration caused by astrocyte reaction are dependent on TAK1 signaling [86]. Another study reported by Katura et al. showed that peripheral nerve injury induces an increase in TAK1 expression in astrocytes in the spinal dorsal horn, and this TAK1 upregulation increases JNK1 phosphorylation in spinal astrocytes and contributes to the development and maintenance of mechanical hypersensitivity [87].

Tissue type plasminogen activator (tPA) is a well-known extracellular serine protease that converts zymogen plasminogen into an active serine protease. tPA is found on the endothelial cells of blood vessels and is involved in the degradation of blood clots [88]. In addition, tPA participates in modification of the extracellular matrix that leads to long-term potentiation in the hippocampus [89]. Kozai et al. reported that tPA is upregulated in spinal astrocytes following root injury [90]. Moreover, continuous intrathecal administration of a tPA inhibitor suppresses root ligation-induced mechanical allodynia. These data suggest that astrocyte-derived tPA in the dorsal horn is essential for the mechanical hypersensitivity following root injury.

Other Molecules

N-myc downstream-regulated gene 2 (NDRG2) is a member of the NDRG family and is widely distributed in the CNS but only expressed in astrocytes. NDRG2 members have different functions in cell differentiation, proliferation, and maintenance of cell morphology [91]. Ma et al. found that down-regulation of NDRG2 in spinal astrocytes inhibits their reactivity and reduces nociceptive behaviors in a rat model of spared nerve injury (SNI) [92]. Another study reported by Li et al. also indicated that inhibition of NDRG2 contributes to astrocyte-specific neuroprotection [93].

FGFR3 is a member of the fibroblast growth factor receptor (FGFR) family, which contains four members (FGFR1–4) that mediate FGF signal transduction. FGF/FGFR signaling plays an important role in cell differentiation, neuronal survival, and cell development [83, 84]. MMP-2 is released by astrocytes after the injury and induces activation of IL-1β [83]. Therefore, it would be ideal to target MMP-2 and reduce inflammation after injury. Unfortunately, most available drugs have a non-specific affinity for MMPs and thus induce various side-effects or minimal desired effects. MMP-1 and MMP-9 inhibitors have recently been developed, and yet not much progress has been made in the development of MMP inhibitors [154]. TNF-α is a cytokine that is usually produced by activated microglia and astrocytes. Administration of TNF-α antibody effectively alleviates hyperalgesia [155], indicating the prospect of anti-TNFα therapy in the treatment of chronic pain. An ongoing phase III clinical trial is aimed to test the efficacy of infliximab (a TNFα antagonist) in treating lower-back pain in patients (NCT03704363). Multiple compounds and medicines targeting astrocytes alleviate chronic pain by regulating pro-inflammatory mediators such as TNFα, IL-1β, and CCL2 [156,157,158,159,160]. Of note, besides the neuronal expression of TNFR1 and IL-1R, they are also expressed in astrocytes and microglia and contribute to glial activation [45, 96, 161]. Pharmacological inhibition of TNF-α also attenuates glial activation then relieves chronic pain. However, since a certain amount of TNFR1 or IL-1R is expressed on neurons, blocking TNF-α also affects neuronal function.

In addition, pharmacological inhibition of P2X3 in rats following CCI of the trigeminal infraorbital nerve attenuates facial pain [82]. The purinergic receptor P2X3 is found in astrocytes in the spinal trigeminal nucleus, and blocking P2X3 inhibits reactive astrogliosis and the release of downstream inflammatory factors [82]. Thus, it might also be possible to target downstream molecules of reactive astrogliosis to reduce its effect on pain behaviors. For example, patients with neuromyelitis optica have been treated with tocilizumab, an IL-6 antibody, which was found to be safe and effective [162]. Although some of these compounds have antinociceptive effects in animal models and inhibit the reactivity of astrocytes, more designs for clinical trials to test their analgesic efficacy on humans are needed.

Another potential approach is to block gap junction proteins, such as connexin-43 (Cx43) and pannexin 1 (Panx1). Cx43 is specifically upregulated in spinal astrocytes after CCI [30]. Once upregulated, Cx43 enhances ATP release from astrocytes and finally leads to microglial activation and allodynia [163, 164]. Besides, Cx43 also contributes to the release of glutamate and chemokines, and blocking this protein remarkably attenuates neuropathic pain sensitization [30, 165]. The main mechanism is that, in chronic pain states, Cx43 modulates hemichannel function, leading to an increase in the permeability to various cytokines and chemokines [30, 166]. In addition, glial Panx1 contributes to the tactile hypersensitivity in chronic orofacial pain by inducing hyper-responsiveness to ATP [167]. Some Panx1 blockers (including mefloquine and probenecid) have been reported to improve morphine withdrawal syndrome [168]. However, further studies are needed to clarify their effects on humans.

Conclusion and Future Perspectives

In summary, we have reviewed different kinds of evidence to demonstrate the necessity and sufficiency of astrocytes in chronic pain. We also explain how astrocytes promote chronic pain through astrocyte-microglia or astrocyte-neuron interactions (Fig. 1). When peripheral nerve injury or tissue damage occurs, astrocytes change to a reactive state in response to different neurotransmitters or neuromodulators in the spinal cord or brain. Reactive astrocytes are usually accompanied by the activation of a variety of intracellular signaling pathways. Therefore, the molecular mechanisms of astrocyte-microglia-neuron crosstalk in the spinal cord and brain under chronic pain conditions deserve further study.

Fig. 1
figure 1

Astrocytic, microglial, and neuronal interaction in chronic pain. Nerve injury induces the release of CXCL13, which activates astrocytes via the CXCR5 receptor. The activation of astrocytes results in the upregulation of CX43 expression and a switch in CX43-mediated function from gap-junction communication to CX43 hemichannel-mediated paracrine signaling, resulting in the increased release of pro-inflammatory cytokines, chemokines, glutamate, and ATP, which activate microglia through P2RX4, P2RX7, and other receptors. The activation of these microglial receptors induces the release of pro-inflammatory cytokines (including TNF-α and IL-1β) and further amplifies neuronal excitability. These cytokines also result in the upregulation of the transcriptional regulators TRAF6, STAT3, and subsequent activation of the JNK and ERK pathways in astrocytes, further increasing their production and release of chemokines and facilitating neuropathic pain. The figure was created with BioRender.com.

Given the important role of astrocytes in the facilitation of chronic pain, targeting them may provide novel prevention and treatment strategies. However, because astrocytes play an essential supportive and protective role in the CNS, it is important to target specific signaling events in astrocytes without disrupting their overall well-being. The recent identification of astrocyte-expressing genes by transcriptome analyses suggests that astrocytes display inter- or intra-regional heterogeneity and act as a gate for descending noradrenergic control of mechanosensory behavior, which indicates the diverse functions and phenotypes of astrocytes for chronic pain regulation [169]. Akdemir et al. also found that subpopulations of Lfng-labeled astrocytes in laminae III and IV of the dorsal horn are involved in the regulation of neuronal activity and maintaining sensory-processing circuity associated with light touch [170]. Ablation of Lfng+ astrocytes reduces glutamatergic synapses and mechanosensory responses. In addition, compared to the classical neuronal gate control theory of pain, Xu et al. reported a new function of astrocytes in the gating of nociceptive signals in the spinal cord [171]. Spinal astrocytes are activated by electrical stimulation of peripheral Aβ fibers, which induces long-term depression in NK1R+ neurons and antinociception. Meanwhile, suppression of reactive astrocytes by different methods blocks such processes. Their results demonstrate astrocytes as a new and important component of pain gating by activation of Aβ fibers that exert non-neuronal control of pain. Thus, these recent discoveries may provide a new research direction for astrocyte regulation of chronic pain.