Purinergic signaling in oligodendrocyte development and function

Abstract

Myelin, an insulating membrane that enables rapid action potential propagation, is an essential component of an efficient, functional vertebrate nervous system. Oligodendrocytes, the myelinating glia of the central nervous system (CNS), produce myelin throughout the CNS, which requires continuous proliferation, migration and differentiation of oligodendrocyte progenitor cells (OPC). Because myelination is essential for efficient neurotransmission, researchers hypothesize that neuronal signals may regulate the cascade of events necessary for this process. The ability of oligodendrocytes and OPCs to detect and respond to neuronal activity is becoming increasingly appreciated, although the specific signals involved are still a matter of debate. Recent evidence from multiple studies points to purinergic signaling as a potential regulator of oligodendrocyte development and differentiation. Adenosine triphosphate (ATP) and its derivatives are potent signaling ligands with receptors expressed on many populations of cells in the nervous system, including cells of the oligodendrocyte lineage. Release of ATP into the extracellular space can initiate a multitude of signaling events, and these downstream signals are specific to the particular purinergic receptor (or receptors) expressed, and whether enzymes are present to hydrolyze ATP to its derivatives adenosine diphosphate (ADP) and adenosine, each of which can activate their own unique downstream signaling cascades. This review will introduce purinergic signaling in the CNS and discuss evidence for its effects on oligodendrocyte proliferation, differentiation and myelination. We will review sources of extracellular purines in the nervous system and how changes in purinergic receptor expression may be coupled to oligodendrocyte differentiation. We will also briefly discuss purinergic signaling in injury and diseases of the CNS.

Description for graphical abstract

Oligodendrocytes are an essential component of a functional vertebrate nervous system. Purinergic receptors regulate oligodendrocyte development, including the proliferation, migration, and differentiation of oligodendrocyte progenitor cells. Extracellular adenosine triphosphate and its derivatives, adenosine diphosphate and adenosine, activate purinergic receptors, each with unique downstream signals. This review introduces purinergic signaling and discusses its effects on oligodendrocyte proliferation, differentiation and myelination.

Introduction

Oligodendrocytes perform the crucial function of synthesizing myelin sheaths to insulate CNS axons and ensure rapid action potential propagation. Myelin production and remodeling continues through adulthood, and the synthesis of new myelin sheaths even plays a role in learning, and differentiation of new oligodendrocytes (OL) from oligodendrocyte progenitor cells (OPC) is an essential part of this process (Gibson et al., 2014). The close physical interaction between neurons/axons and OLs has prompted researchers to wonder if neurons somehow communicate with and influence OLs. However, it is unclear if only the cells that physically contact each other can communicate, or if neurons can signal to OPCs at a distance, potentially influencing their differentiation into myelinating OLs.

Several classic studies demonstrate that glial cells respond to electrically active neurons: tetrodotoxin (TTX) injections into the eye results in decreased OPC proliferation in the optic nerve (Barres and Raff, 1993), optic nerve glia respond via calcium signaling to electrically stimulated axons (Kriegler and Chiu, 1993) and similarly, terminal Schwann cells at the frog neuromuscular junction (NMJ) exhibit calcium transients in response to frog motor nerve stimulation (Jahromi et al., 1992; Reist and Smith, 1992). These studies beg the question as to whether there is a common mechanism mediating all of these glial behaviors and what the signal responsible for these glial responses might be. With the knowledge that ATP packaged in synaptic vesicles acts as a co-transmitter at many neuronal synapses and that adenosine can be released axonally upon stimulation, researchers began testing the hypothesis that purinergic signaling might regulate the development and function of myelinating glia (Abbracchio et al., 2009; Kuperman et al., 1964). These early observations that glia respond to neuronal activity sparked a flurry of interest in activity-dependent regulation of myelinating glia. It is now well appreciated that increases or decreases in neuronal activity affect OL differentiation and myelin production both in vitro and in vivo, in a variety of experimental models (see Almeida and Lyons, 2017; Fields, 2015, for recent reviews). This review will focus specifically on how purinergic signaling affects the development and function of myelinating glia in the CNS.

Types of purinergic receptors and signal transduction mechanisms

ATP and its derivatives, ADP, adenosine monophosphate (AMP) and adenosine, have long been known to be important in cell metabolism. However, the discovery of membrane receptors for extracellular purines, and the acknowledgement that they act as intracellular signaling ligands, opened the door to a new field of cell communication. Members of the purinergic receptor family have nearly ubiquitous expression in many different tissues, including the nervous system, where they regulate processes as diverse as inflammation, vasodilation, and neurotransmission (Burnstock and Knight, 2004). In the CNS, neurons and multiple types of glia are in close proximity, so that a signal such as ATP released by a neuron could exert effects on many nearby cells (Figure 1). Furthermore, even a ligand as ubiquitous as ATP can exert cell- and tissue-specific effects depending on its concentration, the particular purinergic receptors expressed, and whether ectonucleotidase enzymes are present to break down ATP into its derivatives. Figure 1 illustrates purinergic communication between many CNS cell types, and Table 1 summarizes the purinergic receptors expressed by OPCs and OLs. Neurons, astrocytes, and microglia also express purinergic receptors, and cell- and region-specific expression of the many receptor subtypes has been reviewed elsewhere (Burnstock, 2003; Illes et al., 2012; Koizumi et al., 2013; Verkhrasky et al., 2009). Although various ectonucleotidases are ubiquitously expressed throughout the CNS, cell-type specific expression of each subtype has not been thoroughly investigated (Zimmermann, 2006). The complexity and specificity of purinergic signaling in different cell and tissue types is due in part to the number of different kinds of purinergic receptors.

Figure 1. Overview of purinergic signaling in CNS cells.

Schematic of CNS neuron/glial interactions and purinergic receptor expression. It should be noted for neurons and glial cells; expression of receptor subtypes varies across brain regions and with development. More detailed expression data for OPCs and OLs is given in Tables 1 and 2. For simplicity, microglia, which also release and respond to ATP, are not shown. Boxed regions A & B represent the regions of more detailed information found in Figure 2.

Table 1. Expression of purinergic receptors.

	Expression-OPCs	Expression-mature OLs
A1	Y (RTPCR)1 Y (RNAseq)15	Y (RNAseq)15
A2a	Y (RTPCR)1 N (RNAseq)15	N (RNAseq)15
A2b	Y (RTPCR)1 Y (RNAseq)15	N (RNAseq)15
A3	Y (RTPCR)1 N (RNAseq)15	N (RNAseq)15
P2X1	Y (WB)4 N (RNAseq)15	N (RNAseq)15
P2X2	Y (WB)4 N (RNAseq)15	N (RNAseq)15
P2X3	Y (WB)4 N (RNAseq)15	N (RNAseq)15
P2X4	Y (WB)4 Y (RNAseq)15	Y (RNAseq)15
P2X7	Y (IHC)3,17 (WB)4,17 Y (RNAseq)15	Y (IHC)16 N (RNAseq)15
P2Y1	Y (WB)4 (IHC)4 Y (RNAseq)15	Y (IHC)9 N (RNAseq)15
P2Y2	Y (WB)4 Y (RNAseq)15	N (RNAseq)15
P2Y4	Y (WB)4 N (RNAseq)15	N (RNAseq)15
P2Y12	N (WB)4 N (IHC)10 Y (RNAseq)15	Y (IHC)10,12 Y (RNAseq)15
P2Y13	Y (RNAseq)15	N (RNAseq)15

P2X₅, P2X₆, P2Y₆, P2Y₁₁ and P2Y₁₄ are not detected in OL lineage cells.

Y, expression detected; N, expression examined and not detected.

References:

¹Stevens et al., 2002,

²Othman et al., 2003,

³Feng et al., 2015,

⁴Agresti et al., 2005,

⁵Coppi et al., 2013,

⁶James and Butt, 2001,

⁷Kastritsis and McCarthy, 1993,

⁸Kirischuk et al., 1995,

⁹Moran-Jimenez and Matute, 2000,

¹⁰Dennis et al., 2012,

¹¹González-Fernández et al., 2014,

¹²Amadio et al., 2006,

¹³Takeda et al., 1995,

¹⁴He et al., 1996,

¹⁵Zhang et al., 2014 (for RNAseq data, less than 2 FPKM was considered no expression),

¹⁶Matute et al., 2007,

¹⁷Wang et al., 2009.

For all tables, the same reference list and numbering. is used.

Purinergic receptors are divided into two main classes: P1 receptors, which are activated by adenosine, and P2 receptors, which are activated by ATP and/or ADP (Abbracchio et al., 2006). P2 receptors are further sub-divided into P2X or P2Y subtypes, according to whether the receptor is ionotropic (P2X) or metabotropic (P2Y). Four mammalian P1 receptors have been identified, A1, A2a, A2b and A3, all of which are G protein coupled receptors. A1 and A3 inhibit adenylate cyclase, decreasing intracellular cAMP, whereas A2a and A2b activate adenylate cyclase, increasing cAMP, although additional downstream signaling mechanisms, such as mitogen-activated protein kinases and IP₃, have also been identified (Fredholm et al., 2000; Schulte and Fredholm, 2003). In the CNS, P1 signaling has been implicated in regulating neuronal firing in a variety of circuits, in addition to its effects on myelinating glia (Chen et al., 2014; Sebastiao and Ribeiro, 2015; Stevens et al., 2002, 2004).

Metabotropic P2Y receptors are also G-protein coupled receptors, exhibiting the typical seven transmembrane domain structure shared by all G-protein coupled receptors. To date, 8 P2YRs have been identified. P2Y₁, P2Y₂, P2Y₄, P2Y₆, and P2Y₁₁ activate phospholipase C and utilize IP₃ to mobilize internal calcium, whereas P2Y₁₂, P2Y₁₃ and P2Y₁₄ inhibit adenylate cyclase and/or modulate ion channels (Abbracchio et al., 2009, 2006; Verkhratsky, 2005). P2Y₂, P2Y₄, and P2Y₁₁ are preferentially activated by ATP, whereas P2Y₁₂ and P2Y₁₃ are preferentially activated by ADP (Zimmermann, 2006a).

The seven P2X receptor proteins (P2X_1–7), are activated by ATP and form cation channels as homo- or hetero-multimers (Burnstock, 2003). It is now well appreciated that various purinergic receptors are expressed by many types of neurons and glia, and purines can be involved in processes ranging from sleep, myelination, and pain transmission (Figure 1) (Dunwiddie and Masino, 2001; Fields and Stevens, 2000). Because of the often overlapping expression of multiple receptor types in neurons and glia, it has been difficult to work out precisely how ATP and other extracellular purines mediate these effects (Abbracchio et al., 2009; Fields and Burnstock, 2006; Lecca et al., 2012). Experiments utilizing conditional receptor knockouts are now beginning to answer some of these questions.

Sources of purines in the nervous system

All cells synthesize ATP as part of their metabolism and, therefore, are capable of releasing large quantities into the extracellular space if damaged (Casano et al., 2016; Davalos et al., 2005; Elliott et al., 2009). Additionally, multiple mechanisms exist for the regulated release of ATP during normal nervous system physiology, and release of ATP via synaptic vesicles is one well-characterized mechanism (Figure 2A). Most, if not all, neurons use ATP transporters to concentrate neurotransmitters into synaptic vesicles (Abbracchio et al., 2009). ATP is therefore released in high concentrations (up to 1000 mM) as a co-transmitter with neurotransmitters such as glutamate and acetylcholine (Fields and Burnstock, 2006). In some PNS and CNS neurons, ATP is even released as a bona fide neurotransmitter (Burnstock, 1972; Edwards et al., 1992; Holton and Holton, 1954). One study has also reported activity-dependent release of adenosine directly from neurons, although it is unclear whether synaptic vesicles or some other mechanism is the route of release (Wall and Dale, 2007). This neuronal activity-dependent release of ATP at synapses is one way that glial cells could detect neuronal firing and respond with changes in their own signaling or differentiation.

Figure 2. Purine release and signaling in the CNS.

(A) Sources of extracellular purines and (B) expression of purinergic receptors by OPCs/OL. Synaptic terminals, axons, and astrocytes all release ATP into the extracellular space. Expression of ectonucleotidases is ubiquitous, but heterogeneous, in CNS cells, and subcellular localization of ectonucleotidases is not known. For simplicity, purinergic receptors are only shown on OPCs and OLs.

However, axon terminals are not the only sites of ATP release in the nervous system. Multiple studies demonstrate ATP release from axonal segments remote from synaptic terminals or from cultured neurons that haven’t formed any synapses (Figure 2B) (Edstrom et al., 1992; Fields and Ni, 2010; Kriegler and Chiu, 1993; Stevens and Fields, 2000). This ATP may come from “extrasynaptic” vesicles released from premyelinated/unmyelinated segments of axons (Figure 2B). Studies in mice and more recently, zebrafish, demonstrate the accumulation of synaptic vesicles along axonal domains in contact with OPC processes (Bergles et al., 2000; Koudelka et al., 2016; Lin and Bergles, 2004) and ultrastructural analysis previously revealed features of presynaptic and postsynaptic specializations on axonal and OPC membranes, respectively (Bergles et al., 2000). Although these studies did not test whether these vesicles contained ATP, it is likely that ATP would be released as a co-transmitter in the same way it is at axon terminals. Recently, the neurotransmitters glutamate and GABA have been observed to affect OPC migration and differentiation (reviewed in Gallo et al., 2008). These observations have supported the idea that activity-dependent signals from neurons may communicate with OPCs and regulate the process of myelination. Since ATP is also released from axons and can signal to OPCs and OLs, purinergic signaling may be an additional activity-dependent signal involved in regulating myelination. Evidence supporting this idea will be reviewed in detail in the section “The effects of purinergic signaling on OPC proliferation, migration, differentiation and myelination.”

Another proposed mechanism for axonal ATP release is via volume activated anion channels (VAACs) (Figure 2B). These channels can be activated by cell swelling and/or mechanical stress and are a well-characterized mechanism of ATP release in non-neuronal cells (Burnstock, 1999; Sabirov and Okada, 2005). One in vitro study demonstrated that axonal swelling and mechanical stress as a result of firing action potentials can activate VAACs, leading to ATP leaking out of VAACs (Fields and Ni, 2010). Whether VAACs are involved in axonal ATP release in vivo remains to be determined. We also note that VAAC-mediated versus vesicular release of ATP from axons may not necessarily be mutually exclusive. Both of these mechanisms center on activity-dependent release from pre-myelinated and/or unmyelinated axons, which could potentially couple myelinating glial proliferation, differentiation, and or myelination with patterns of neuronal firing. It is unclear whether myelinated axonal segments continue to release ATP, and whether this has any effect on the myelinating glial cell associated with the axonal segment.

In addition to activity-dependent release, there is also evidence for ATP release via ATP-binding cassette transporters and pannexin channels on neurons and glia (Figure 2B). Additionally, astrocytes can release ATP via vesicular release or through gap junction hemichannels (Abbracchio et al., 2009; Boué-Grabot and Pankratov, 2017).

After release, extracellular ATP binds to and activates P2 receptors or is hydrolyzed by extracellular enzymes. Figure 3 summarizes release mechanisms for ATP and the extracellular enzymes involved in its breakdown. These mechanisms rapidly clear ATP from the extracellular space while also creating new ligands, ADP and adenosine (Figures 2&3). The production of ADP and adenosine by ectoenzymes can have synergistic or opposing effects, depending on the cell type and receptors present. The extracellular enzymes ectonucleotide pyrophosphatase/phosphodiesterase (enpp) and ectonucleoside triphosphate diphosphohydrolase (entpd) hydrolyze ATP to ADP and AMP (Figures 2&3). Nt5e, also known as CD73, is the main enzyme that converts AMP to adenosine (Figures 2&3) (Zimmermann, 2000). It is enriched in OPCs compared to other neural cell types, and is downregulated with differentiation (Zhang et al., 2014). Other ectonucleotidases also exhibit differential expression in OPCs, differentiating OLs, and myelinating OLs, suggesting the possibility that they play a role in regulating OL differentiation (Figure 4). Because OL lineage cells express receptors for ATP, ADP and adenosine, it is possible that differential expression of ectonucleotidases regulates the effects of extracellular purines on OL migration and differentiation. The enzyme Enpp2, also known as Autotaxin, promotes OL differentiation and myelination, but this function is mainly attributed to effects on lysophosphatidic acid signaling and extracellular matrix remodeling, rather than its nucleotidase enzymatic activity (Dennis et al., 2005; Fox et al., 2003; Wheeler et al., 2015). Interestingly, one study demonstrated a functional interaction between P2Y12 and Enpp2 in differentiating OLs (Dennis et al., 2012). The function of ectonucleotidases in nervous system development is unclear, although there is some evidence for their involvement in regulating neuronal excitability (Carlsen and Perrier, 2014). The role of ectonucleotidases in regulating purinergic signaling between neurons and OLs is a promising area for future study.

Figure 3. Purinergic release, receptors, and degradation.

ATP can be released into the extracellular space via membrane channels or exocytosis. In neurons, ATP is released in high concentrations from synaptic vesicles. Extracellular enzymes rapidly degrade ATP into ADP, AMP, and adenosine. ATP and ADP activate metabotropic P2Y and/or ionotropic P2X receptors. Adenosine activates P1 receptors.

Figure 4. Developmentally regulated expression of purinergic receptors and ectonucleotidases in the OL lineage.

Because few studies directly compare purinergic protein expression, the majority of data for this figure comes from the RNAseq database of Zhang et al., 2014. A comprehensive list of expression data for purinergic receptors in OPCs and mature OLs can be found in Table 1. It should be noted that OL lineage cells may have heterogeneous expression depending on brain region, age, or other factors.

Purinergic receptor expression by OL lineage cells

Kastritsis and McCarthy cultured OPCs and mature oligodendrocytes from neonatal rat cortex and tested their response to various neurotransmitters. They observed calcium signaling in both OPCs and OLs when treated with ATP and the P2 agonists 2MeSATP and gammaSATP, demonstrating that both OPCs and OLs express functional P2 receptors (Kastritsis and McCarthy, 1993). Kriegler and Chiu also demonstrated calcium responses in intact optic nerve (P2 and P7) preps treated with ATP and adenosine, although they did not distinguish between OPCs, OLs and other glia in their experiments (Kriegler and Chiu, 1993). Since then, attempts at elucidating the contributions of specific purinergic receptors to OL-lineage cell functions has been a complicated task, especially given that ATP is rapidly converted to ADP and adenosine in vivo and in vitro, and each ligand activates receptors which may initiate unique downstream signaling events (Jacobson and Müller, 2016). Furthermore, selective agonists and antagonists for individual P2 receptors are lacking, and a number of papers have mistakenly used 2MeSATP as a selective agonist for P2Y over P2X receptors, when in fact it activates most P2X receptors at the low micromolar (and higher) concentrations used in most studies (Jacobson and Müller, 2016). In vitro, studies have reported expression of all four P1 receptors and many P2 receptors by OL-lineage cells (summarized in Table 1) (Agresti et al., 2005; Othman et al., 2003; Stevens et al., 2002). However, there is some discrepancy over purinergic receptor expression in OL lineage cells, as in vivo and in vitro expression data has had some contradictory results (see Table 1). For example, P2Y12 was not detected in cultured OPCs by Western blot, but another study detected mRNA for P2Y12 using RNAseq (Agresti et al., 2005; Zhang et al., 2014).

The differences observed in expression among various studies may arise, in part, from the different brain regions being examined and the age of the animal used. Whereas many in vitro studies cultured OPCs from P1 mouse cortex, Zhang, Chen, and Sloan et al. used freshly dissociated P17 mouse cortex. Amadio et al. observed P2Y12 expressed in mature OLs in rat cerebral cortex, striatum and peduncles, but not other brain regions, so it is possible that OL lineage cells are heterogeneous in their expression of and response to purinergic receptors (Amadio et al., 2006). In support of this, recent RNAseq data has identified at least 13 subpopulations of OL lineage cells based on expression profiles (Marques et al., 2016).

Developmental changes in receptor expression

Previous studies report heterogeneity in the responses of OL-lineage cells to ATP depending on factors like the age of the animals or the brain regions being analyzed (Bernstein et al., 1996; He et al., 1996). For example, cultured OLs derived from adult rat spinal cord did not respond at all to ATP, whereas OLs derived from P21 rat spinal cord showed calcium signaling in response to ATP (He et al., 1996; Takeda et al., 1995). Another study examined OPCs in corpus callosum slices taken from animals at different ages. A high percentage of OPCs responded to ATP in younger animals (P3–P18), whereas significantly fewer OPCs from older animals (P11–P18) responded (Bernstein et al., 1996). However, the receptor(s) mediating this variability in calcium responses were not identified. It is important to note that in these comparisons, differentiated OLs from young animals are also responsive to ATP, so the effect of age is separate from differentiation state of OPCs. These observed differences in the OPC response to ATP in young versus adult animals suggests potentially different functions and characteristics for “adult OPCs,” also known as NG2 cells. It is also unclear whether purinergic signaling may have different roles in white vs gray matter OLs.

How differentiation affects purinergic receptor expression in OL lineage cells has been the subject of some controversy. In one study, a calcium response was observed in differentiated OLs treated with ATP, but not in OPCs (Kirischuk et al., 1995). In contrast, other studies observed calcium signaling in OPCs and differentiating OLs treated with ATP or synthetic P2 agonists (Agresti et al., 2005; Kastritsis and McCarthy, 1993; Stevens et al., 2002). In P17 mouse cortex, mRNA expression for P2RX7, P2RY1, and P2RY12 is decreased in differentiating and myelinating OLs compared to OPCs, suggesting that responsiveness to ATP might be downregulated with differentiation (Zhang et al., 2014). However, another study showed increased expression of P2Y₁₂ protein in differentiating OLs (Dennis et al., 2012). Figure 4 summarizes the current literature on expression changes of purinergic receptors and ectonucleotidases in OPCs, pre-myelinating OLs, and myelinating OLs. Since the age of animals, brain regions used, and culture conditions can all effect how OLs respond to various purines, care must be taken when comparing and interpreting these experiments. It is even possible that whether OLs are cultured with axons or in monoculture alters their response to ATP (He et al., 1996). Clearly, more studies are needed directly comparing P2 expression and calcium signaling in OPCs and differentiated OLs in different brain regions to help elucidate the role of purinergic signaling in these cell populations.

The effects of purinergic signaling on OPC proliferation, migration, differentiation and myelination

OLs have an essential role in efficient neurotransmission in the CNS. Myelin sheaths produced by differentiated OLs wrap and insulate axon segments, and small, electrically conductive nodes of Ranvier are spaced between myelin internodes. This arrangement allows rapid saltatory conduction of action potentials along myelinated axons and is crucial for proper nervous system development and function and impaired myelination has been implicated in neuropsychiatric disorders and cognitive disabilities (Haroutunian et al., 2014). Furthermore, the extensive physical interaction between OLs and axons allows for OLs to be involved in maintaining neuronal homeostasis. OLs assist with axonal energy metabolism by exporting lactate, an energy substrate; through myelin membranes (see Saab et al., 2013 for review).

In addition to OLs, OPCs also contact axonal segments in the CNS. Ultrastructural studies have revealed synapse-like structures between axons and OPCs, with synaptic vesicles in the axon and structures resembling postsynaptic densities in the OPC process (Bergles et al., 2000). More recently, live, in vivo imaging has demonstrated synaptic vesicle accumulation along axonal segments during contact with OPC processes (Hines et al., 2015). In this study, synaptic vesicle release along unmyelinated axon segments promoted myelin sheath formation. Growing evidence points to neuronal activity and neurotransmitter release affecting OPC proliferation, migration, and differentiation, as well as myelin formation (Almeida and Lyons, 2017; Fields, 2015; Gallo et al., 2008). Furthermore, continuous myelin remodeling throughout adulthood points to a need for continuous communication between neurons, OPCs, and OLs (Sampaio-Baptista and Johansen-Berg, 2017). Since ATP is released from neurons in an activity-dependent manner (see above), ATP (or its derivatives ADP and adenosine) may be involved in the ongoing communication between neurons and myelinating glia.

Evidence for the role of purinergic signaling in regulating OPCs, OLs, and myelination comes largely from in vitro studies. Testing the hypothesis, that axonally-released adenosine mediated OPC proliferation and calcium signaling, Stevens et al. demonstrated calcium responses in OPCs either via electrical stimulation of co-cultured neurons, or by directly applying the P1 agonist NECA to OPC monocultures (Stevens et al., 2002). Additionally, P1 antagonists blocked the activity-dependent calcium response; supporting the conclusion, that adenosine released from firing neurons stimulates calcium signaling in OPCs. Adenosine and the P1 agonist NECA also inhibited proliferation and promoted differentiation and myelin production, whereas the P2 agonist 2MesATP did not affect differentiation (Stevens et al., 2002). The conclusions from cell culture are supported by evidence in cerebellar slices that adenosine inhibits proliferation and increases differentiation of OPCs (Stevens et al., 2002). More recently, in vitro pharmacology experiments tested the roles of specific P1 and P2 agonists, but with some conflicting results. For example, multiple studies have reported that adenosine inhibits OPC proliferation (Agresti et al., 2005; Stevens et al., 2002). However, selective agonists for A1 and A2a receptors did not have any effect on OPC proliferation (Coppi et al., 2013; Othman et al., 2003). In another example of conflicting results, one study found that the A1 agonist CPA promoted OPC migration in a dose-dependent manner, whereas another study reported that neither adenosine nor CPA had any effect on OPC migration (Agresti et al., 2005; Othman et al., 2003). Stevens et al. observed that adenosine promotes differentiation of OPCs and myelination (Stevens et al., 2002). However, the A2a agonist CGS21680 was reported to inhibit differentiation, and the A1 agonist CPA had no effect on OPC differentiation (Coppi et al., 2013; Othman et al., 2003). These experiments are summarized in Tables 2 and 3. Because ligands selective for a single P2 receptor subtype are lacking, and most P2 ligands activate or inhibit more than one receptor type, we have organized the effects of P2 receptors on OL proliferation, migration, and differentiation according to ligand, rather than by receptor (Jacobson and Müller, 2016).

Table 2. P1 receptor expression and function in OL lineage cells.

	Expression-OPCs	Expression-mature OLs	Migration	Proliferation	Differentiation
A1	Y (RTPCR)1 Y (IHC)2 Y (RNAseq)15	Y (RTPCR)11 Y (IHC)2 Y (RNAseq)15	(+)2 no effect4	No effect2	No effect2
A2a	Y (RTPCR)1 N (RNAseq)15	Y (RTPCR)11 N (RNAseq)15		No effect5	(−) 5
A2b	Y (RTPCR)1 Y (RNAseq)15	Y (RTPCR)11 N (RNAseq)15
A3	Y (RTPCR)1 N (RNAseq)15	Y (RTPCR)11 N (RNAseq)15
P1 (unspecified)	n/a	n/a	No effect4	(−) 1,4	(+) 1

Y, expression detected; N, expression examined and not detected; (+) increased, (−) decreased.

References:

¹Stevens et al., 2002,

²Othman et al., 2003,

³Feng et al., 2015,

⁴Agresti et al., 2005,

⁵Coppi et al., 2013,

⁶James and Butt, 2001,

⁷Kastritsis and McCarthy, 1993,

⁸Kirischuk et al., 1995,

⁹Moran-Jimenez and Matute, 2000,

¹⁰Dennis et al., 2012,

¹¹González-Fernández et al., 2014,

¹²Amadio et al., 2006,

¹³Takeda et al., 1995,

¹⁴He et al., 1996,

¹⁵Zhang et al., 2014 (for RNAseq data, less than 2 FPKM was considered no expression),

¹⁶Matute et al., 2007,

¹⁷Wang et al., 2009.

For all tables, the same reference list and numbering. is used.

Table 3. Effects of purinergic receptor ligands on OPCs and mature OLs.

	Ca2+ response-OPC	Ca2+ response-mature OL	Migration	Proliferation	Differentiation
Adenosine		Y6 N8	No effect4	(−)1,4	(+)1
ADP	Y4	Y6,8	(+)4	(−)4
ATP	Y4,7 N8	Y6,7,8 Y/N13,14	(+)3,4	(−)4
BzATP	Y4		(+)3 No effect4
ADPβS	Y4		(+)4	(−)4	(+)4
2MeSADP
2MeSATP	Y1	Y1,6,13		No effect1	No effect1
ATPγS	Y1,4			(−)4 No effect1
α,β-MeATP	N4	Y6 N8		No effect4

Y Ca₂⁺ response detected, N response not detected, Y/N response dependent on brain region or culture conditions; (+) increased, (−) decreased.

References:

¹Stevens et al., 2002,

²Othman et al., 2003,

³Feng et al., 2015,

⁴Agresti et al., 2005,

⁵Coppi et al., 2013,

⁶James and Butt, 2001,

⁷Kastritsis and McCarthy, 1993,

⁸Kirischuk et al., 1995,

⁹Moran-Jimenez and Matute, 2000,

¹⁰Dennis et al., 2012,

¹¹González-Fernández et al., 2014,

¹²Amadio et al., 2006,

¹³Takeda et al., 1995,

¹⁴He et al., 1996,

¹⁵Zhang et al., 2014 (for RNAseq data, less than 2 FPKM was considered no expression),

¹⁶Matute et al., 2007,

¹⁷Wang et al., 2009.

For all tables, the same reference list and numbering is used.

Purinergic signaling in white matter injury and disease

ATP and adenosine are well-known modulators of immune function, acting as both activators and chemotactic signals for a variety of immune cells (Bao et al., 2013; Chimote et al., 2013; Honda et al., 2001; Wong and Schlichter, 2014). Along these lines, there is growing evidence for purinergic signaling contributing to pathology and degeneration in white matter diseases and traumatic injury, but it is still unclear to what extent purinergic signaling affects OLs directly (Rivera et al., 2016; Rivkees and Wendler, 2011; Turner et al., 2002, 2003). During injury and degeneration, massive amounts of ATP are released from damaged cells. This ATP can lead directly to excitotoxic death of nearby cells and increase inflammation, both of which lead to further degeneration (Matute, 2011). The evidence for purinergic modulation of immune cells in various injury and disease models is extensive (see Beamer et al., 2016; Idzko et al., 2014; Jacob et al., 2013 for recent reviews). In this section, we will focus on studies involving white matter injury or disease and discuss evidence for purinergic signaling on OPCs and OLs.

P2X7 is one purinergic receptor that seems to directly contribute to OL death and myelin loss in white matter pathology. This cation channel is expressed by many nervous system cells, including OPCs and OLs (Agresti et al., 2005; Matute et al., 2007). Unlike other P2X channels, P2X7 receptors do not desensitize with prolonged activation. The sustained, high levels of extracellular ATP that occur with excitotoxicity, ischemia, traumatic injury, and degeneration lead to pathophysiological, prolonged activation of P2X7 receptors and calcium influx in OLs and other cells (Skaper et al., 2010). With prolonged activation, P2X7 receptors form large-diameter, nonspecific pores, which in combination with high calcium influx, are thought to contribute to cell death (Matute, 2011). Studies using cultured OLs and intact optic nerves have observed that high concentrations of ATP or the P2X7 agonist BzATP result in OL death, which can be prevented with P2X7 antagonists (Matute et al., 2007; Wang et al., 2009). In vivo, there is evidence for P2X7 activation contributing to OL death in models of spinal cord injury, ischemia, and models of demyelinating diseases (Domercq et al., 2010; Matute et al., 2007; Wang et al., 2009, 2004). In fact, P2X7 antagonists protect against OL cell death in models of ischemia (Domercq et al., 2010; Wang et al., 2009). However, since many CNS cells and peripheral immune cells express P2X7 receptors, it is impossible to say whether the beneficial effects of P2X7 antagonists in vivo are due to direct effects on OLs.

Purinergic signaling has also been studied in the context of demyelinating and degenerative diseases. Amyotrophic lateral sclerosis (ALS) is a progressively degenerative disease primarily affecting motor neurons. Although motor neuron degeneration is the hallmark of ALS, post-mortem tissue samples have revealed myelin abnormalities and OL degeneration in ALS patients (Philips et al., 2013). Apoptotic and dysmorphic OLs are observed before the onset of motor neuron death in an animal model of ALS (Kang et al., 2013). There is recently some evidence of a purinergic contribution to ALS. The adenosine receptors A1 and A2a are known to modulate motor neuron firing, and these receptors have been investigated as therapeutic targets to slow disease progression (Volonté et al., 2016). P2X7 is also believed to be involved in disease onset and progression, potentially contributing to reactive gliosis and activation of microglia. However, there is some controversy on the role of P2X7, as conflicting results have been obtained using either antagonists to P2X7 or genetic knockout (see Volonté et al., 2016). Despite promising evidence of purinergic signaling in ALS pathogenesis, and the recent appreciation of OL pathology as a contributing factor, the effect of purinergic signaling on OLs in the context of ALS has not yet been investigated. Because of the growing evidence for purinergic signaling affecting OLs in development and disease, we believe this is a promising area for future study.

Multiple sclerosis (MS) is a demyelinating disease with an unknown etiology and a complex interplay of inflammation, myelin damage and OL loss. Although the initial insult contributing to myelin degeneration is unknown, high levels of ATP released from damaged tissue appear to contribute to pathological immune activation and progressive myelin loss. Because animal models of MS that have been used to study purinergic signaling typically employ global genetic knockouts or systemic drug administration, it is difficult to elucidate effects of purinergic receptor antagonists or knockouts directly on OLs versus on immune cells. Demyelination, OL death, and other symptoms of experimental autoimmune encephalomyelitis (EAE), a well-established model for MS, were ameliorated by treatment with P2X7 antagonists (Matute et al., 2007). Given the evidence for toxic P2X7 activation contributing to OL death in injury models (see above), a direct protective effect of the P2X7 antagonist on OLs is likely, although immune involvement certainly can’t be ruled out. A gain-of-function mutation in P2RX7 that results in increased cytotoxic calcium influx has even been identified in MS patients (Oyanguren-Desez et al., 2011). Increased P2X7 expression in the CNS in EAE and in MS patients compared to healthy controls has also been observed (Amadio et al., 2017; Grygorowicz et al., 2010). However, there is some controversy over P2X7 and MS, since separate studies using P2X7^−/− null mice reported contradictory results. In one case, P2X7 knockout exacerbated EAE pathology (Chen and Brosnan, 2006). In the other study, P2X7 knockouts had greatly reduced incidence of EAE, although the animals that did develop the disease exhibited similar symptom severity to WT animals (Sharp et al., 2008). Because of the ubiquitous nature of purinergic signaling and the complexity of EAE/MS pathology, it is difficult to draw definitive conclusions on precisely how P2X7 signaling may be affecting disease onset and progression, but these questions certainly warrant future study.

Whether P2X7 receptors expressed by OL-lineage cells play any role during normal physiology is unknown. It is hypothesized that other P2X receptors may act as neuronal activity detectors, since ATP levels could reach high enough concentrations to briefly activate these receptors during bursts of neuronal firing. However, since P2X7 receptors require particularly high concentrations of ATP to become activated, they may act as early damage detectors for myelin (Matute, 2011). Another possibility is that P2X7 activation may recruit OPCs to remyelinate damaged tissue, since a recent study demonstrated that P2X7 activation promoted OPC migration in vitro (Feng et al., 2015). In addition to P2X7, changes in OPC P2Y2 and OL P2Y12 receptor expression have been observed in models of demyelination, but how these changes could be contributing to or protecting from demyelination is unclear (Amadio et al., 2010; Moyon et al., 2015).

In contrast to the pro-inflammatory and neurodegenerative effects attributed to ATP, adenosine has generally been thought of as protective in models of white matter disease and injury. For example, a number of studies demonstrate that A2a null mice have exacerbated EAE pathology, but this effect is attributed to the anti-inflammatory effects of adenosine on peripheral immune cells and microglia (Ingwersen et al., 2016; Yao et al., 2012). Effects of purinergic signaling specifically on OLs have not been explored in this model. Similarly, A2a activity was found to protect white matter in models of spinal cord injury by exerting anti-inflammatory effects on infiltrating immune cells (Genovese et al., 2009; Li et al., 2006). The A1a adenosine receptor also appears to have a protective effect in EAE, as A1a^−/− null mice developed more severe EAE than WT mice (Tsutsui et al., 2004). Furthermore, in a model of toxin-induced demyelination, agonists for A1 receptors promoted remyelination (Asghari et al., 2013). Altogether, these models demonstrate that adenosine appears to exert protective effects on myelin and/or OLs, although the mechanisms may involve immune modulation rather than direct effects on OLs. Paradoxically, however, A3 agonists contribute to OL cell death, and inhibiting adenosine signaling through A2a receptors can also be protective in models of white matter injury (González-Fernández et al., 2014; Li et al., 2006; Melani et al., 2009). This discrepancy is thought to be due to opposing actions of adenosine receptors expressed by peripheral versus CNS cells. It is still unknown in any of these models whether adenosine directly affects OL-lineage cells. Understanding potentially opposing effects of adenosine on OLs and immune cells in disease pathology is of critical importance when considering potential therapeutics. Because there is ample evidence for purinergic signaling regulating OL differentiation and myelin production during development, there is a strong likelihood for these signals to also affect myelin repair in disease contexts. The availability of conditional adenosine receptor knockouts now allows for the potential to answer some of these questions.

Conclusions

OPCs and OLs express many purinergic receptors, some of which seem to be developmentally regulated. Multiple studies have demonstrated that extracellular purines can influence the proliferation, migration and/or differentiation of OPCs into OLs and it is clear that purinergic signaling plays a role in regulating OLs in the CNS. Because ATP release from neurons correlates with neuronal activity, it is possible that purinergic signaling orchestrates interactions between neurons and OPCs/OLs in order to fine-tune myelin development. The ubiquitous nature of ATP release from neurons, combined with developmental differences and potential heterogeneity of receptor expression in OL-lineage cells provides the potential for highly context-dependent responses to ATP release depending on brain region, age, and even levels of activity. The many different purinergic receptors, the lack of selective ligands, and the ubiquitous presence of extracellular enzymes that convert ATP to ADP, AMP, and adenosine, make interpretations of any experiments based on pharmacology difficult. This complexity may underlie the sometimes-conflicting results in the current literature. However, this complexity is also, what makes purinergic signaling such a powerful and nuanced regulator of a heterogeneous cell lineage, which has different functions depending on differentiation state and brain region. We believe that the current studies have merely scratched the surface of how extracellular purines regulate neuron-OL interactions, and care must be taken when making direct comparisons between studies that may be using OLs in very different contexts. Careful characterization of receptor expression and genetic knockout experiments are needed to more precisely determine the roles of each receptor in OL development.

Abbreviations

2MeSATP

2-Methylthioadenosine triphosphate is a potent P2Y purinoceptor agonist

ALS

amyotrophic lateral sclerosis, also known as Lou Gehrig’s Disease

ATP

adenosine triphosphate

ADP

adenosine diphosphate

AMP

adenosine monophosphate

BzATP

2′(3′)-O-(4-Benzoylbenzoyl)adenosine-5′-triphosphate tri(triethylammonium) salt, P2X7 agonist and P2X1/P2Y1 partial agonist

cAMP

cyclic adenosine monophosphate

CGS21680

adenosine A2 receptor agonist

CNS

central nervous system

CPA

N6-cyclopentyladenosine, a A1 selective agonist

EAE

experimental autoimmune encephalomyelitis

Enpp2

Ectonucleotide Pyrophosphatase/Phosphodiesterase 2, also known as Autotaxin

Entpd

ectonucleoside triphosphate diphosphohydrolase

gammaSATP

adenosine triphosphate gamma S. Substrate and inhibitor of ATP-dependent enzyme systems. Hydrolyzed very slowly by phosphatases and most ATPases. Once thiophosphorylated, proteins are resistant to protein phosphatases. P2 purinergic receptor agonist.

IP₃

Inositol trisphosphate (IP3), used for signal transduction in biological cells

MS

multiple sclerosis

NG2

Neuron-glial antigen 2, a chondroitin sulfate proteoglycan that in humans is encoded by the CSPG4 gene. A subset of oligodendrocyte progenitor cells are called NG2 cells.

NECA

5′-(N-ethylcarboxyamido)adenosine, an adenosine receptor agonist

NMJ

neuromuscular junction

Nt5e

5′-Nucleotidase Ecto, also known as CD73. Catalyzes the conversion of extracellular nucleotides to membrane-permeable nucleosides.

OL

oligodendrocyte

OPC

oligodendrocyte progenitor cells

TTX

tetrodotoxin

A1, A2a, A2b, A3 Receptors

adenosine receptors

P2X Receptors

ligand-gated ion channel family that open in response to extracellular ATP

P2Y Receptors

G-protein-coupled receptors that respond to extracellular purine and pyrimidine nucleotides

VAAC

volume-activated anion channels