Adenine nucleotides (AdNs) play important roles in immunity and inflammation. Extracellular AdNs, such as adenosine triphosphate (ATP) or nicotinamide adenine dinucleotide (NAD) and their metabolites, act as paracrine messengers by fine-tuning both pro- and anti-inflammatory processes. Moreover, intracellular AdNs derived from ATP or NAD play important roles in many cells of the immune system, including T lymphocytes, macrophages, neutrophils and others. These intracellular AdNs are signaling molecules that transduce incoming signals into meaningful cellular responses, e.g. activation of immune responses against pathogens.
Adenine nucleotides (AdNs) are well-known endogenous small molecules important in energy metabolism. In particular, adenosine triphosphate (ATP) is the major molecule produced intracellularly to store energy released by the catabolism of nutrients, such as carbohydrates, lipids, amino acids and others (Figure 1). Another important small molecule in the process of cellular catabolism is nicotinamide adenine dinucleotide (NAD). This dinucleotide serves as a co-substrate for oxidoreductases, thereby transferring hydrogen atoms into the central process of ATP synthesis, oxidative phosphorylation, that takes place in mitochondria (Figure 1).
Adenine nucleotides as paracrine mediators and intracellular second messengers of immunity and inflammation.
Besides their role in energy metabolism, ATP and NAD are also substrates for enzymatic synthesis of extra- and intracellular signal molecules, known as first and second messengers.
Both ATP and NAD can be released into the extracellular space, either as a result of cell lysis or by controlled release processes (reviewed in ref. ). In such ATP- or NAD-rich microenvironments, these molecules may be degraded by specific ectoenzymes to adenosine that acts on purinergic P1 receptors, thereby counteracting inflammation (Figure 1) (reviewed in refs [2,3]). In contrast, ATP acts directly on purinergic P2 receptors that enhance inflammation (reviewed in ref. ). For NAD, different scenarios were described. In the presence of ADP-ribosyl transferases, it may act in a pro-inflammatory fashion by activation of P2X7 receptor via ADP-ribosylation of Arg125 . On the opposite, very recently, NAD was reported to signal via P2Y receptors, thereby suppressing ATP-evoked release of the pro-inflammatory cytokine IL-1β , revealing an anti-apoptotic effect of NAD. These processes may also proceed in parallel, depending on the expression of ectoenzymes and/or receptors in the microenvironment (Figure 1).
In the cytosol, ATP is converted by adenylyl cyclases into the second messenger 3′,5′-cyclic AMP (cAMP) that further acts on protein kinase A (PKA) or exchange protein directly activated by cAMP (Epac) for signal transduction (Figure 1). NAD is a substrate for NAD glycohydrolase/ADP-ribosyl cyclase (NADase/ADPRC) CD38 to produce second messengers such as adenosine diphosphoribose (ADPR), 2′-deoxy-ADPR (2dADPR), or cyclic adenosine diphosphoribose (cADPR) (reviewed in refs [7,8]) (Figure 1). While the latter acts on ryanodine receptors (RYR), likely type 2 and/or 3 RYR , ADPR and its 2′-deoxy derivative modulate open probability of the cation channel transient receptor potential, subtype melastatin 2 (TRPM2)  (Figure 1). Nicotinic acid adenine dinucleotide phosphate (NAADP) is also formed from NAD via NADP, either by CD38 or by a so far unknown ‘NAADP enzyme’ (reviewed in ref. ). Targets reported for NAADP are RYR1 or two-pore channels (TPCs), likely modulated by an NAADP-binding protein (NAADP-BP) (reviewed in ) (Figure 1).
Following, we review specific tissues where the role of AdN in inflammation is currently emerging, e.g. adipose tissue or the CNS; these first two chapters are more meant as introductory overview. Furthermore, we discuss recent advances and novel findings regarding second messengers such as cAMP, ADPR/2dADPR and NAADP in inflammatory responses. Since an important aspect of AdN research are AdN derivatives, such as inactive and/or membrane-permeant prodrugs that can be used in intact cells or whole animals, the last chapter is devoted to chemical approaches towards prodrugs for AdN intracellular messengers.
Purinergic signaling in brown fat/inflammation
White adipocytes are the major constituents of white adipose tissue (WAT) and store energy-dense triglycerides that are released as fatty acids during catabolic conditions . Beige adipocytes that emerge in specific WAT depots in response to various conditions such as cold exposure, as well as brown adipocytes, the defining parenchymal cells of brown adipose tissue (BAT), internalize triglycerides that can be combusted for heat production to maintain body temperature . Although the metabolic functions of WAT and BAT are primarily performed by adipocytes, tissue-resident endothelial cells, macrophages and immune cells have important roles to maintain metabolic homeostasis . The relevance of this cellular interaction becomes apparent in obese individuals, as here disturbances in the intercellular communication trigger a chronic inflammatory response in adipose tissues, which contributes to the development of insulin resistance, hyperlipidemia and type 2 diabetes . Recent studies suggest that extracellular AdNs regulate metabolic and differentiation processes in WAT and BAT. For example, it was shown that ATP-induced inflammation in white adipose tissue drives tissue-resident Th17 cells in metabolically unhealthy obesity and that ATP acting via P2X7 promotes a Th17-polarizing microenvironment in WAT . In addition, CD39 is involved in insulin regulation and hepatic metabolism , whereas CD73 plays a role in intracellular lipid metabolism in muscle and adipose tissue [18,19]. Further studies have implicated extracellular purinergic signaling by adenosine as an important mechanism for the thermogenic response mediated by brown and beige adipocytes . Little is known about the autocrine and paracrine regulation controlling the interaction between macrophages, T cells or endothelial cells with thermogenic adipocytes. It is well established that these immune cells are equipped with a variety of receptors for extracellular nucleotides, including P2X receptors and nucleotide-metabolizing ectoenzymes such as CD39 . Notably, a key role for P2X7 in mediating fever in mice after injection of lipopolysaccharide (LPS) was demonstrated . LPS-induced fever is known to involve BAT activation . Moreover, genetic deletion of P2X7 or specific antagonism of P2X7 using small molecule compounds decreased the febrile response to LPS . Altogether, these data suggest a role of extracellular purinergic signaling for the interplay between thermogenic adipocytes, endothelial and tissue-resident immune cells in BAT, which may be relevant for the regulation of metabolic and thermogenic responses. However, the role of extracellular AdN receptors and enzymes for paracrine signaling, which may be relevant for lipid uptake and thermogenic responses, is unclear.
Purinergic signaling in astrocytes in neuroinflammation
Astrocytes are the most abundant glial cells in the mammalian brain. They fulfill many functions in development and physiology such as control of neurite growth, metabolic homeostasis, neurovascular coupling and synaptic modulation . Astrocytes express a variety of purinoceptors and respond to ATP and ADP via P2 receptors, but also to adenosine via P1 receptors . The tissue-specific pattern of ATP release sites, astrocytic purinoceptors, ATP-degrading ectoenzymes and purine transporters allows astrocytes to accomplish different tasks that correspond to the demands of the cellular environment and brain region. In addition, astrocytes change the expression pattern of purinoceptors under pathological conditions and evolve from metabolically supportive to immunocompetent cells  during the inflammatory response after epileptic seizures and stroke, e.g. astrocytes undergo reactive gliosis, which involves (but is not restricted to) cell proliferation, migration, elongation of cell processes and increased expression of glial fibrillary acidic protein [28,29]. While physiological stimulation of astrocytes by low concentrations of ATP, as obtained by the synaptic release of ATP, activates the high-affinity P2Y1 and P2X1 receptors , reactive astrogliosis is elicited by high concentrations of ATP that occur after neuronal cell damage and activate the low-affinity P2X7 receptor in addition to the high-affinity receptors . Furthermore, neuroinflammation causes astrocytes to increase the expression of P2X7 receptors and hence renders them even more susceptible to ATP . Reactive astrocytes release pro-inflammatory factors such as IL1-β, TNF-α and INF-γ . In addition, the release of ATP from astrocytes through hemichannels is enhanced during neuroinflammation due to an increased expression of connexins and recruitment of additional hemichannels . High concentrations of ATP released from reactive astrocytes recruit adjacent astrocytes and microglial cells, which, in turn, acquire a pro-inflammatory phenotype themselves . Astrocyte-derived ATP and cytokines may also affect endothelial cells, as discussed above for BAT, resulting in compromised blood–brain barrier integrity and penetration of the blood–brain barrier by immune cells such as T lymphocytes . In addition to the pro-inflammatory response of astrocytes upon purinergic stimulation, anti-inflammatory responses have also been found in reactive astrocytes. Ischemic tolerance mediated by reactive astrocytes is enhanced by stimulation of P2X7 receptors and subsequent up-regulation of hypoxia-induced factor 1 . Anti-inflammatory cytokines IL-10 and TGF-β are released from reactive astrocytes in inflammatory tissue of animals with experimental autoimmune encephalomyelitis, resulting in reduced inducible nitric oxide synthase activity, improved blood–brain barrier integrity and a less severe neuroinflammatory response . However, it is not known in detail until now, which circumstances favor an anti- versus a pro-inflammatory phenotype of reactive astrocytes upon purinergic stimulation.
TRPM2 and ADPR/2dADPR in inflammation
TRPM2 is expressed in most innate immune cells  as well as dendritic  and effector T cells . Its extended C-terminus harbors a domain homologous to NUDT9, an ADPR pyrophosphatase. ADPR, the substrate of NUDT9, activates the channel in a Ca2+- [40,41] and temperature-  dependent manner.
Activation of TRPM2 occurs downstream of reactive oxygen species (ROS) and inflammatory mediators like LPS , TNF-α  and fMLP [45–47] contributing to the production of chemokines and pro-inflammatory cytokines by monocytes and macrophages, chemotaxis [45,47] and neutrophil extracellular trap formation  by neutrophils, maturation and migration of dendritic cells , and proliferation and secretion of cytokines by effector T cells . TRPM2 also limits oxidative stress during inflammation by depolarizing the plasma membrane thereby counteracting NADPH oxidase activity  and by up-regulating heme oxygenase 1 .
Large amounts of ADPR are produced when DNA-strand breaks are inflicted by exposure of cells to ROS, leading to activation of poly-ADP ribose polymerase 1 (PARP-1) which results in poly-ADP-ribosylation of nuclear proteins . Hydrolysis of poly-ADP ribose by poly-ADP ribose glycohydrolase then releases ADPR that may activate TRPM2, e.g. during reperfusion after ischemia .
Different models have been proposed for the activation of TRPM2 under physiological conditions, including direct oxidation of the channel by ROS  and activation by Ca2+ [54,55], but many recent publications indicate an important role of CD38 in the activation of TRPM2 [56–59].
Besides hydrolyzing NAD to ADPR, CD38 can also turn 2′-deoxy-NAD into the highly efficient and less Ca2+-sensitive TRPM2 agonist 2dADPR . CD38 was initially introduced as ecto-NAD glycohydrolase. Since it does not only produce the TRPM2 agonists ADPR and 2′-deoxy-ADPR, but also the second messenger cADPR, De Flora et al. (reviewed in ref. ) coined the expression of the ‘topological paradoxon’, meaning that for an ectoenzyme to produce intracellularly active second messenger molecules, transport processes across the plasma membrane for both substrate(s) and product(s) must exist in cells. However, later it was found that a fraction of CD38 is expressed in type III orientation  and could thus produce both ADPR and 2dADPR at the cytoplasmic face of the membrane. CD38 activity has previously been shown to be regulated by redox state and oxidative stress ; consistently, the increase in 2dADPR in T cells in response to hydrogen peroxide is CD38-dependent .
The relative importance of ADPR and 2dADPR for signaling by inflammatory mediators and whether activation of CD38 only occurs downstream of intracellular ROS production or if type III CD38 is also regulated by post-translational modification clearly deserves further investigation.
cAMP in inflammation
cAMP, as an ubiquitous second messenger, plays a crucial role for immune control and inflammation. In particular, cAMP actions in T cells are considered to convey pronounced immunosuppressive and anti-inflammatory effects. Increase in cAMP levels in T cells, by stimulating cAMP synthesis with, e.g. prostaglandin or adenosine (the latter acts via P1 receptors), or inhibiting its degradation, causes a marked immunosuppression and T cell anergy, i.e. a state of reduced function in which a viable, antigen-specific T cell is unable to respond to an immunogenic stimulus . Interestingly, kinetics and amplitude of cAMP signaling may modulate Ca2+ signaling, as one of the essential pathways of T cell activation differentially: while transient and low elevations of cAMP support subsequent signaling via the T cell receptor/CD3 complex, high and long-lasting cAMP concentrations antagonize signaling via the T cell receptor/CD3 complex .
Importantly, cAMP is the central component of regulatory T cell (Treg)-mediated immunosuppression, as its increased formation in Treg is a prerequisite for their suppressive capacity . Moreover, direct cAMP transfer between Treg and conventional T cells has been proposed to occur via gap junctions and to suppress various immune responses . Although this process has been implicated in the suppression of conventional T cell activation and graft-versus-host disease , no direct imaging of cAMP transfer between Treg and conventional T cells has been performed. Alternatively, extracellular nucleotides ADP and ATP can be converted to adenosine by CD39 and CD73 expressed on the Treg cell surface. Adenosine generated in proximity of conventional T cells can activate P1 receptors on their surface to stimulate intracellular cAMP production. The latter paracrine mechanism might be important, especially under conditions of inflammation-associated hypoxia . However, such kind of cAMP cross-talk between two T cell types has also not been directly visualized in pairs of cells, nor have the source of high cAMP levels in Treg and the mechanisms of their differential regulation in Treg vs. Teff cells been systematically explored. These lines of research would be exciting to be followed in future studies.
NAADP and Ca2+ microdomains in T cells during inflammatory processes
As pointed out above, T cells are of utmost importance for adaptive immune responses. Upon processed antigen recognition, an immune synapse between antigen-presenting cell and T cell is formed. The first intracellular signals observed in T cells in such immune synapses are localized, small and transient Ca2+ signals, termed Ca2+ microdomains. Such Ca2+ microdomains occur within tens to hundreds of milliseconds. A prime candidate for triggering Ca2+ microdomains is NAADP, since endogenous NAADP is rapidly formed in T lymphocytes and other cell types in response to different stimuli [67–69]. Then, NAADP likely binds to a specific, but so far molecularly non-identified NAADP-BP [70–72]. Together, NAADP and NAADP-BP then provide an initiating Ca2+ signal by activation of a Ca2+ release channel on Ca2+ storing organelles . In T cells, there is accumulating evidence that the Ca2+ channel involved is RYR1 (for details, see paragraph below). In some other cell types, evidence for TPCs as NAADP-sensitive channels was presented [74–76], though in other studies TPCs were characterized as Na+ or H+ channels, modulated mainly by phosphatidylinositol 3,5-bisphosphate [77,78]. However, a common theme in NAADP signaling is amplification of an NAADP-dependent primary, local Ca2+ signal by cADPR- or d-myo-inositol 1,4,5-trisphosphate (IP3)-dependent mechanisms via Ca2+-induced Ca2+ release . Upon microinjection of NAADP into Jurkat T lymphoma cells, Ca2+ microdomains were observed within tens of milliseconds; gene silencing of RYR abolished effects of NAADP, while microinjection of IP3 was not affected . Site-directed stimulation of the T-cell receptor/CD3 complex plus co-stimulatory receptor CD28 in murine primary T cells resulted in a highly dynamic process of Ca2+ microdomain formation that started very close to the site of activation . In RYR1−/− T cells, the number of Ca2+ microdomains was markedly and significantly decreased, suggesting that RYR1, likely activated by NAADP in the context of an NAADP-BP, is a major Ca2+ source for Ca2+ microdomains in T cells . Furthermore, a Ca2+ entry component is of similar importance for Ca2+ microdomain formation and was very recently identified as ORAI1, activated by both adaptor proteins stromal interaction molecules (STIM) 1 and 2 .
Taken together, the highly dynamic process of Ca2+ microdomain formation in T cells involves as major mechanisms NAADP and its target, RYR1, closely localized to and cooperating with ORAI1 and STIM1/2.
Synthesis of AdN for immunity/inflammation research
As detailed above, a better understanding of the NAADP/Ca2+-signaling system is of interest as this dinucleotide is the most powerful Ca2+-mobilizing agent known to date. The same holds true for 2dADPR, the most potent activator of TPRM2, or the role of cNMP signaling in the context of inflammation and regulation of immune response (see chapters above). All these endogenous AdNs constitute a research field of rising interest.
However, cell-based studies with these compounds are very difficult to perform because of their high polarity which prevents an uptake into cells and therefore the application of these compounds is carried out via electroporation, microinjection or infusion via patch-clamp pipettes.
To overcome these problems, two approaches involving chemical modification of these second messengers, e.g. NAADP, were reported. A photo-labile moiety was used and the resulting biochemically inactive, photocaged derivative expected to enable the release of NAADP in dependence of a light pulse. However, the compound still required to be applied intracellularly by microinjection . Another approach aimed at synthesizing a lipophilic NAADP derivative by random attachment of acetoxymethyl groups . A study performed by Yu et al.  demonstrated that cell-permeable ADPR analogs are useful tools to investigate TRPM2 mechanisms in more detail. They reported the synthesis of 1-(2-nitrophenyl)ethyl (NPE)-caged ADPR derivatives. However, two compounds were isolated in low yields, one bearing a single NPE group while the other compound comprised five NPE groups. Only the former compound could be photo-chemically cleaved to form ADPR. First reports on protected cNMPs included cAMP and cGMP derivatives carrying benzyl groups at the phosphate . Later, also the synthesis of compounds following the acetoxymethyl strategy was reported . However, in almost all cases, the synthesis was found to be very difficult and sometimes even the characterization of the material was doubtful. Therefore, the stepwise, convergent synthesis of membrane permeable, bio-reversibly modified and/or photocaged derivatives of second messengers is still highly desirable. First approaches using the acyloxybenzyl strategy introduced by us in the field of antivirally active nucleoside di- and triphosphate prodrugs  showed promising results .
In summary, AdNs play important roles in inflammation and immunity in two ways: as paracrine mediators allowing for cell-to-cell communication in inflammatory (micro)environment and as second messengers regulating diverse intracellular functions to orchestrate adequate physiological responses of single (immune) cells within multicellular organisms. As pointed out above, many aspects of AdN biology and biochemistry are still unknown; membrane-permeant prodrugs will, together with many other experimental approaches, help to clarify these points in the near future.
2′-deoxy adenosine diphosphoribose
brown adipose tissue
cyclic adenosine diphosphoribose
cyclic nucleoside monophosphates
glial fibrillary acidic protein
nicotinic acid adenine dinucleotide phosphate
nicotinamide adenine dinucleotide
reactive oxygen species
stromal interaction molecules
transient receptor potential, subtype melastatin 2
white adipose tissue
The authors are jointly funded by the Deutsche Forschungsgemeinschaft in the frame of Collaborative Research Center (SFB) 1328 ‘Adenine Nucleotides in Immunity and Inflammation’ [Projektnummer 335447717-SFB 1328, projects A01 (A.H.G), A04 (C.M.), A05 (R.F.), A06 (V.O.N.), A07 (C.L.) and A10 (J.H. and F.K.-N.)].
We are grateful to our coworkers for intense laboratory work and fruitful discussions.
The Authors declare that there are no competing interests associated with the manuscript.
All authors contributed equally to this work.