The Importance of NAD in Multiple Sclerosis

2.2. Biosynthesis of NAD

Twentieth century man is particularly susceptible to dietary induced NAD deficiency. The most devastating nutritional deficiency disease in the history of the United States of America was the NAD deficiency disease pellagra, an epidemic which killed over 120,000 people in the first two decades of the 1900s [5]. NAD precursors are distinguished as one of the few known molecules identified as important enough to be forced into the public diet for improved health by the U.S. government [6]. It should be realized however, that niacin requirements of individuals can vary depending on genetics and stress. Significantly, disease pathology is well known to actively promote NAD depletion through poly(ADP)ribose polymerase (PARP-1) activation. Recently, new NAD metabolic pathways have gained recognition for their important roles in MS. For example, the NAD precursor nicotinamide (NA) can ameliorate MS in the experimental autoimmune encephalomyelitis (EAE) animal model [7]. Thus, we focus here on NAD biosynthesis to understand the potential role of NAD depletion in MS pathogenesis.

In all vertebrates, NAD can be synthesized by two pathways: de novo synthesis from tryptophan [3, 4] and/or from vitamin precursors in the diet: NA, nicotinamide (NAM), and nicotinamide riboside (NAMR). The pathway from tryptophan to the NAD precursor nicotinic acid mononucleotide (NaMN) requires four enzymes encoded by the genes KMO, KYNO, HAAO, and QPRT. NaMN is then converted to NAD via three reactions catalyzed by NARPRT, NMNAT, and NADS. For the de novo pathway to be completed it is necessary to have adequate riboflavin (vitamin B2), pyridoxyl phosphate (vitamin B6), and ascorbate (vitamin C). Deficiencies in any of these can lead to the accumulation of “kynurenine pathway” intermediates (Fig. 1). NAD can also be synthesized from three different essential non-amino acid precursors in the diet: nicotinic acid (NA), nicotinamide (NAM), and nicotinamide riboside (NAMR). Yeast with an inactivated de novo pathway may use any of the salvageable precursors (NA, NAM, NAMR, or nicotinic acid riboside) to support life [8]. Nicotinic acid riboside has only been examined in yeast. These non-amino acid dietary NAD precursors are collectively termed vitamin B3 and require two (NAM, NAMR) or three (NA) steps respectively to generate NAD [1].

Open in a separate window

Fig. (1)

The pathway from NAD precursors to NAD biosynthesis is shown with molecular structures and abbreviations used throughout the manuscript. Also shown, are the reactions that use NAD as a substrate, not a co-factor. Products of ADP cyclase and Sirtuin catalyzed reactions generate products causing release of calcium. CD38 is involved in the persistent release of intracellular calcium connected with hematopoietic chemotaxis and microglial activation, both of which are likely to be of great importance in MS pathogenesis. Feedback inhibition loops are drawn with heavy lines. Additional co-factors required for completion of de novo pathway NAD biosynthesis include ascorbate (vitamin C), pyridoxyl phosphate (vitamin B6), and riboflavin (vitamin B2). Of likely significance to nicotinic acid distinguished capacity as a provider of NAD, NAPRT is not inhibited by NAD feedback inhibition (see text for reference), while NAMPT is inhibited by NAD.

It should be realized that not every cell is capable of converting each precursor to NAD at all times although NAD is essential to survival. Because tissue and cell-type specific enzyme expression differences exist in metazoans, the precursors are differentially utilized in the gut, brain, blood and other organs [1] (Table 1). The de novo pathway is clearly active in liver, neuronal and immune cells. The pathway from NA is expressed in the liver, kidney, heart and intestine. While, vertebrates do not possess the nicotinamidase enzyme that converts NAM to NA, intestinal bacteria in the vertebrate gut use nicotinamidase to convert NAM to NA, which in turn is used to synthesize NAD via the Preiss and Handler salvage pathway. The NAMR salvage pathway is expressed in neurons and in cardiac and skeletal muscle [1].

Collectively, experimental observations support the idea that by increasing NAD, we may be able to prevent or reduce the effects of MS pathogenesis. Mammals elevate their NAD during autoimmune encephalomyelitis [7, 9], while tryptophan levels are decreased in the cerebrospinal fluid (CSF) and sera of MS patients [10]. Caloric restriction (CR) raises NAD levels [11] and provides protection against EAE-mediated pathogenesis [12-14]. CR reduces inflammation, demyelination, and neurodegeneration, but does not suppress immune function. CR increases both the expression of the rate-limiting enzyme controlling NAD biosynthesis [15] and lifespan in a SIRT1 dependent manner [16]. Consequently, the body may synthesize NAD to help control autoimmune encephalomyelitis.

The enzymes required to initiate de novo NAD biosyn-thesis are nicotinamide phosphoribosyltransferase (NAMPT), nicotinic acid phosphoribosyltransferase (NAPRT), indole-amine 2,3-dioxygenase (IDO/INDOL and IDO2/INDOL2), or tryptophan dioxygenase (TDO2). The NAD salvage pathway recycles NAM back to NAD starting with the rate-limiting enzyme NAMPT [17]. The NAD recycling pathway responds to stress conditions in a highly evolutionarily conserved manner, by increasing expression of nicotinamidase (PNC) in prokaryotes or by increasing NAMPT in vertebrates (for a review [18]).

NAMPT requires the substrate NAM, which is rapidly generated following PARP-1 activation in response to genomic damage. PARP-1 activity may also change in direct interferon-γ (IFNγ) causes robust induction of IDO activity in response to local changes in NAD concentration-an active area of research investigation in vivo [19]. Although inflammation-mediated PARP-1 activation can lead to a rapid NAD depletion, NAD can be quickly restored through the NAD salvage pathway. The NAD salvage pathway recycles NA to NAD. Induction of NAMPT or increased NMNAT in Wld^S dramatically increases NAD levels during times of PARP-1 activation. Other distinct pathways to increase levels of tissue NAD include activation of either TDO2 or IDO. While TDO2 has traditionally been understood to restricted in expression in the liver, it is in fact also expressed in neurons and glia [20]. IDO expression is highly induced in professional APCs and neurons by the immune system. IDO is also detectable in endothelial cells [21]. In neurons interferon-γ (IFNγ) causes robust induction of IDO activity in primary human neurons and a slight reduction of TDO2 activity [22]. Although NAD levels were not measured in these experiments, similar IFNγ induction of IDO has been shown to increase NAD levels in macrophages [23, 24] and glia [25, 26]. It is unclear whether IFNγ directly increased NAD in neurons. Other experiments suggest that the tryptophan de novo biosynthesis pathway is not efficient enough to provide neuroprotection [27]. By contrast, other mononucleotide containing precursors in the salvage pathway including NaMN or nicotinamide mononucleotide (NMN) are neuroprotective in the Wallerian degeneration assay.

2.3. Changes in NAD Levels in Animal Models of MS and NAD Precursors

Under conditions of NAD deficiency, neurons are exceptionally vulnerable to the degeneration characteristic of MS [28]. The EAE model of MS indicates that endogenous levels of NAD are elevated in the CNS [7, 9]. This presumably occurs due to the combined increases in lymphocyte infiltration, hematopoietic cell proliferation, and IDO activity in professional APCs commonly observed in MS.

While net increases in NAD in the CNS are observed in EAE models, we predict that these increases are largely restricted to the immune system and come at the risk of NAD deficiency in neurons. Extracellular levels of the precursor used for de novo biosynthesis (tryptophan) are significantly decreased in the serum and CSF of MS patients [10]. Fortunately, this loss of available NAD precursor can be partially rescued either by administrating pharmacological doses of nicotinamide or by enforced caloric restriction [11]. Both approaches lead to increased NAD levels and profoundly ameliorate EAE pathogenesis [7, 12-14]. By contrast, administration of tryptophan is known to increase lymphoproliferation [29, 30]. Thus high dose tryptophan administration may be expected to exacerbate immune-mediated demyelinating pathogenesis, actually promoting neurodegeneration due to hyper activation of the immune system (discussed in greater detail in section 8.1.1). Similarly, inhibition of IDO, which can increase tryptophan levels, exacerbates EAE pathogenesis [31-33]. The latest research indicates that stem cell mediated prevention and recovery from EAE requires IDO activity [34]. Thus it is important to limit the NAD available to lymphocytes but not at the expense of starving neighboring cells of NAD.

3.2. Clinical Course of MS

MS can be characterized by either episodic acute periods of worsening (relapses, exacerbations, bouts, attacks), gradual progressive deterioration of neurologic function, or combinations of both. The clinical course of MS is classified into four forms: relapsing-remitting (RR), primary progressive (PP), secondary progressive (SP), and progressive relapsing (PR) [56]. RR-MS is defined by disease relapses with full recovery or with sequelae. PP-MS progresses continuously from the onset. Initial RR disease is often followed by progression (SP-MS).

Clinical and immunological profiles can be different among different forms of MS or even among the same form of MS. However, in general, attack by inflammatory cells in the CNS results in disease onset, exacerbation/relapse and disease progression (Table 2), while inhibition of autoimmune reactions lead to disease remission. Demyelination and axonal degeneration result in clinical attack and disease progression. Irreversible axonal damage and lack of remyelination contribute to a lack of recovery. Since remyelination in the CNS, if at all, is a very slow process, recovery from clinical signs should be attributed to subsiding inflammatory responses, but not necessarily neuro-regeneration. Gliosis may further inhibit regeneration processes in the CNS. In section 9 we will later discuss that all three components are connected, and targeting one component can affect the other component in some cases.

3.3. Etiology of MS

Whether infectious agents, environmental toxins, mutagenesis, or nutrient deficiencies cause MS is uncertain. Inheritance studies of MS have found genetic links affecting disease susceptibility, but have not revealed any single genetic mutation [57]. Given the high frequency of MS in the population, it seems most likely that more than one factor is involved. For example, geographical epidemiology and animal models clearly reveal important roles for vitamin D signaling [58], while deficiencies of vitamin B12 [59-61] or sadenosylmethionine [62, 63] can directly cause demyelination. Meanwhile recurrent reactivation of a latent herpes infection resembles relapsing-remitting MS and is commonly found in the CNS, where its potential role in MS pathogenesis is difficult to resolve owing to its nearly ubiquitous presence [64].

The autoimmune etiology of MS is supported by 1) evidence of immune cells infiltrating the CNS, 2) the presence of anti-myelin T cell and antibody responses, 3) a susceptibility associated with major histocompatibility complex (MHC) haplotypes, and 4) altered clinical signs after immunomodulation therapy [52]. Currently, type I interferon (IFNα and β), glatiramer acetate, and antibody against very late antigen (VLA)-4 have been shown to suppress clinical signs of MS.

It has been proposed that MS is mediated by anti-myelin helper T cell type (Th1 or Th2) responses. The Th1 cells produce cytokines such as IL-1, IFNγ, tumor necrosis factor-α/β (TNFα and TNFβ/lymphotoxin-α). These cytokines help are involved in cellular immune responses including delayed type hypersensitivity (DTH). The Th2 cells produce cytokines including interleukin (IL)-4, 5, 6, and 10, which contribute to antibody production. This is partly because a human clinical trial of IFNγ has been reported to exacerbate MS [65, 66]. Although the study was controversial (the observation period, small number of patients, clinical outcome of follow up period, etc), no large clinical trial using IFNγ has been tried since then. In fact, there is some evidence to support that Th1 responses are protective in the early stages of the disease and IFNγ treatment early on can prevent the development of MS [67, 68].

IFNγ clearly exerts a protective role in viral animal models of MS [69]. Perhaps most significantly with respect to NAD metabolism, the IFNγ has been shown to require indoleamine 2,3-dioxygenase (IDO) activity to exert many of its effects including anti-viral (herpes [70, 71] and CMV [72]), anti-parasitic (malaria [73]), anti-bacterial (streptococci [74]), and anti-intracellular pathogen (T. gondii [75, 76] and Chlamydia [77]).

4.2. Amelioration of EAE Pathogenesis in Wld^S Mice

Two independent research groups have tested Wld^S mice in animal models of MS [37, 97, 98]. Tsunoda et al. [98] induced EAE in Wld^S mice and their parent strain, C57BL/6 (B6) mice, using MOG35-55 peptide. Both mouse groups showed a similar disease onset, MOG-specific lymphoproliferative responses in the periphery, and CNS inflammation and demyelination at 2-week post immunization (p.i.) with MOG. Similar levels of peripheral immune responses between two mouse groups were expected, since no immunological deficit has been reported in Wld^S mice. However, Wld^S mice showed complete recovery with a decline in MOG-specific T cell responses. In contrast, wild-type B6 mice showed disease progression clinically with high levels of axonal degeneration, demyelination, and MOG-specific lymphoproliferative responses. Chitnis et al. [97] used identical mouse strains and EAE generating protocols and obtained similar results with Wld^S dramatically delaying EAE disease progression. Further Chitnis et al. noted a reduction in macrophage accumulation and activated microglia for the Wld^S mouse.

Similarly, Kaneko et al. [7] demonstrated an attenuation of clinical signs of MOG35-55-induced EAE in Wld^S mice compared with wild-type B6 mice. Histologically, at 2 weeks post-inoculation (p.i.), B6 mice showed more severe axonal degeneration with increased microglia/macrophage infiltration [97] than Wld^S mice, while no significant difference was seen in the extent of demyelination or CD4+ and CD8+ T cell infiltration in the CNS between the two groups at 2 weeks p.i. In addition, there were no differences between the two mouse groups in MOG induced T cell infiltration, proliferation, delayed type hypersensitivity (DTH) responses, IL-5, 6, 10, and IFNγ production, at 2 weeks p.i., demonstrating that there is no deficit in priming in Wld^S mice [97]. NAD levels were preserved only in Wld^S mice at 2 and 4 weeks, p.i. Daily administration of 500 mg/kg (40g/170lb human equivalent) nicotinamide to EAE mice from the day of MOG immunization or from day 10 pi., reduced all pathology parameters. This included reductions in axonal loss, T cell infiltration and demyelination. NAD levels were quantitatively increased in the spinal cord by the nicotinamide administration. In that study, autoimmune Th1 immune response against MOG was not examined in the nicotinamide treatment group, although others showed that nicotinamide can modulate IFNγ-mediated reactions, and that NAD itself can cause cell death of T cells, particularly CD4+CD25+ regulatory T (Treg) cells [100, 101]. Thus, it is not clear whether or not amelioration of EAE in the nicotinamide treatment group is solely attributed to preservation of axons.

While it is beyond the scope of this review to discuss the role of the Ufd2 aspect of the Wld^S gene, Ufd2 regulates multi-ubiquitination of proteins targeted for degradation by the 26S proteasome complex. Chitnis et al. [97] demonstrated elevated expression of CD200 in neurons and axons in the CNS of naïve Wld^S mice and Wld^S mice with EAE, which was associated with decreased ubiquitination of CD200, most likely caused by the Ufd2 portion of Wld^S gene. Administration of CD200 antibody into Wld^S mice with EAE exacerbated clinical signs of EAE, suggesting that the CD200-CD200R pathway plays a role in attenuating EAE in Wld^S mice. CD200-CD200R interactions play a role in the control of myeloid cellular activity. Interestingly, activation of the CD200 receptor with CD200-Ig in mouse plasmacytoid dendritic cells causes an increase IDO expression and function that exerts a toleragenic effect on autoreactivity [102]. Conversely, IDO inhibition exacerbates EAE pathogenesis prevents stem cell rescue of EAE [31, 33, 34]. Thus, we may expect that CD200-Ig mediated increases in IDO may ameliorate the clinical symptoms of MS. In section 10 we discuss the recent discovery of another soluble factor (TREM-2) that controls IDO expression and has been determined to be present at exceptionally high levels in the cerebrospinal fluid of MS patients [103].

SIRT1 is a member of a highly conserved gene family, the Sirtuins, encoding NAD dependent deacetylases, which deacetylate histones leading to increased DNA stability and prolonged survival in yeast and higher organisms, including mammals. Neuroprotection mediated by SIRT1 activation has been demonstrated in axotomized dorsal root ganglion neurons [36]. Shindler et al. [104] tested whether two SIRT1 activators, nicotinamide riboside (NAMR, SRT647) and resveratrol (SRT501) protects from optic neuritis in EAE mice sensitized with myelin proteolipid protein (PLP)139-151 peptide. Intravitreal injection of NAMR or resveratrol prevents loss of retinal ganglion cells (RGC), but does not prevent inflammation or clinical signs of EAE. The neuroprotective activity was blocked by sirtinol, a SIRT1 inhibitor. Although specific substrates of SIRT1-mediated RGC survival are not known in this study, apoptotic proteins can be candidates for the downstream targets of SIRT1-mediated RGC neuroprotection, since SIRT1 is known to act on proteins involved in apoptosis. In addition, as Bogan et al. [1] proposed, the specific utilization of NAMR by neurons may provide qualitative advantages over niacins in promoting function in the central and peripheral nervous system [1].

In contrast, Suzuki and Koike [105] showed that resveratrol diminished resistance of cerebellar granule cells of Wld^S mice to axonal degeneration induced by colchicine. The authors showed that resveratrol promotes tubulin deacetylation by acting directly on SIRT2, and SIRT2 silencing restores the resistance to axonal degeneration in resveratrol-treated Wld^S neurons.

7.2. IDO

The enzyme indoleamine 2,3-dioxygenase (IDO) mediates the extracellular depletion of the sole endogenous precursor of NAD biosynthesis (tryptophan). Accordingly, serum tryptophan concentrations are consistently low and directly linked to poor prognosis in autoimmune diseases in general starting from the persistent Th1 cytokine activation of IDO [122]. Tryptophan levels are decreased during MS [10, 140] and novel assays are being developed for monitoring immune mediated changes in tryptophan levels for consideration in the context of multiple sclerosis [141]. Activation of IDO exerts the therapeutically beneficial effects of decreasing autoreactive T cell proliferation as well as exerting antimicrobial effects. However, persistent activation of IDO in a conflicted immune system can lead to NAD depletion in otherwise healthy neighboring collateral tissues (reviewed [142]).

Professional APCs including microglia, macrophages, and dendritic cells quickly begin to express large amounts of active IDO when exposed to the Th1-derived cytokines including IFNγ and IFNα. Depletion of extracellular tryptophan by professional APCs causes autoreactive T cells to decrease in number [29, 30]. Dendritic cells are considered the most potent modulator of the adaptive immune system [143]. This is consistent with the exceptionally highly level of IDO inductions commonly seen in dendritic cells. While most typically studied in the context of skin, dendritic cells are also in fact present in the CNS [144] where they are being considered for their potential role in the breakdown of myelin autoantigenicity seen in MS (for a review [145]). Finally, stem cells preventing EAE pathogenesis require inducible IDO function [34]. All observations indicate that professional APCs expressing IDO play a role in controlling multiple sclerosis disease progression.

Similar to professional APCs, neurons exposed to IFNγ also increase IDO expression, while repressing the expression of tryptophan dioxygenase (TDO2) [22]. However, the functional significance of IFNγ-mediated IDO induction and whether NAD levels are actually increased remains to be determined. Both IDO and TDO2 use tryptophan as a substrate to commit to the de novo NAD biosynthetic pathway. Alternatively, when the pathway is not completed, intermediates may be formed. This is known as the kynurenine pathway. Accumulation of these intermediates most frequently occurs during times of co-factor deficiency when nutrition is lacking. IDO activity is highly regulated by cytokines where it primarily functions in regulating the immune system. Meanwhile TDO2 is mostly known for controlling organismal NAD biosynthesis in the liver. In fact however, very little is known with respect to the importance of IDO mediated NAD biosynthesis or extra-hepatic TDO2 function respectively. The functional purpose of this contrasting IFNγ-mediated transcriptional regulation in neurons remains to be determined [22].

EAE models indicate NAD depleting pathways are activated based on detection of increased PAR formation in EAE models [9, 146]. In summation, we expect the concentration of NAD within professional APCs to be elevated during inflammation such that a net increase in NAD is detected, but at likely expense of NAD in other cells, where neurons are well known to be particularly vulnerable to neurodegeneration due to NAD deficiency as witnessed in the classic NAD deficiency disease pellagra [147, 148].

While increased IDO activity causes depletion of the extracellular NAD precursor tryptophan, it can increase intracellular NAD levels. This gives the host immune cells directly combating infectious pathogens (e.g. macrophages), a competitive survival advantage. Interferon causes significant increases in NAD levels in astrocytes [25, 26] or macrophages [23, 24]. Inhibition of IDO [23, 25, 26] prevents IFNγ-mediated increases in NAD; by contrast inhibition of PARP-1 [23, 24] increases IFNα stimulated NAD production in glia and macrophages. Whether or not IFNα can cause an increase in neuronal NAD levels however, is unknown. Many of interferon's activities are completely dependent on the strong interferon-mediated induction of IDO activity [70-77]. Furthermore IFNγ-IDO mediated increases in NAD levels can provide protection against otherwise lethal hydrogen peroxide treatment in glial cells [25], our understanding of the IFNγ-IDO mediated NAD biosynthetic pathway is severely limited.

Interestingly, the addition of the de novo NAD biosynthetic pathway intermediate kynurenine can also actively exert anti-inflammatory effects [149, 150]. It is possible that kynurenine is being used in neighboring cells as a NAD precursor, such that the active IDO-mediated generation of NAD as a second messenger may be more important to immunosuppression than the depletion of tryptophan. Consistent with this idea is the newly appreciated role of IDO-mediated signaling from professional antigen presenting cells that stimulates T regulatory cells to send toleragenic signals to other T cells [150]. Accordingly it may be more the active generation of NAD as a second messenger in professional antigen presenting cells that is more important in immunosuppression than the depletion of NAD precursors locally. Ultimately more research is needed to increase our basic understanding of IDO-mediated immunosuppression.

7.3. CD38

CD38 is an ectozyme with ADP cyclase activity that uses NAD to generate cyclic ADP ribose (cADPR), nicotinic acid dinucleotide phosphate (NAADP), or ADP ribose (ADPR) (for a review [151]). These products then act as potent second messenger activators of calcium channels to control chemotaxis of dendritic cells [152] and activation of microglia [138]. All studies to date indicate that CD38 enzyme activity is constitutive, such that changes in CD38 activity are primarily mediated through transcriptional regulation or subcellular location. CD38 is expressed highly expressed in the brain as well as a variety of blood cells including T cells, B cells, monocytes, red blood cells, and platelets. CD38 is also expressed in the pancreas, prostate, spleen, and heart. While CD157 is CD157 is a related family member, we are limiting this discussion to CD38 since far less is known regarding CD157, but the potentially redundant role of CD157 in controlling NAD levels is completely unknown.

CD38 uses extracellular NAD to generate second messengers that open intracellular calcium stores to control lymphocyte chemotaxis. CD38 exerts significant control over total tissue NAD levels in animal [153-155]. TNFα signaling in macrophages causes a dramatic up-regulation of CD38 that leads to a decrease in NAD levels. This is likely to be a part of TNFα-mediated chemotaxis originating from the Th1 cell that are newly present in MS pathogenesis. Thus, the characteristic CNS lymphoproliferation is likely to be sending signals to microglia via Th1-TNFα binding to TNFα receptor on microglia which through CD38 promotes gliosis (third histological component of MS pathogenesis described in section 2.1). Thus, inhibition of CD38 can be expected to minimize gliosis as well as increase NAD levels. However, this may come at the expense of increased susceptibility to infection. The increased lymphocyte proliferation and infiltration seen in MS is likely to contribute to appreciable losses in neuronal NAD. There are many reasons to consider the function of CD38 as related to MS. First, CD38 is highly expressed in the afflicted and pathogenic tissues of the MS of brain and lymphocytes respectively (for a review [151]). Similarly, CD38 is well established for its likely involvement in possible autoimmune pathogenesis of type II diabetes [156]. Significantly, CD38 appears to be one of the primary regulators of NAD levels in tissue [136, 153-155]. The original defining function of CD38 was determined to lie in controlling innate and adaptive immune functions [152, 157]. Meanwhile, recent studies reveal that by altering NAD levels, CD38 exerts strong control over bioenergetic aspects of metabolism particularly in the context of obesity [155]. Such control over NAD-related bioenergetics may play a role in MS pathogenesis as has described in the bioenergetics section above linking maintenance of NAD levels to maintenance of ATP levels with resistance to EAE-mediated neurodegeneration [37].

The products of CD38 catalyzed degradation of NAD are among the most potent known molecules controlling the release of intracellular calcium stores ([158-160]). These products are cyclic ADP ribose (cADPR) or nicotinic acid adenine dinucleotide phosphate (NAADP) or ADP ribose (ADPR), each of which is produced with differing degrees of reaction efficiency. The CD38 inhibitor dihydroxyazobenzene (DHAB) has been shown to exert cytotrophic effects, keeping cells alive through inhibition of persistent increases in intracellular calcium (Fig. 5; [161]). Microglial cell activation [138] and dendritic cell migration [152] can be controlled by CD38 dependent calcium signaling pathways. All of these observations support the notion that CD38 is highly likely to play important roles in controlling MS pathogenesis. However, no research has been reported describing the potential importance of CD38 in MS. Thus, more research is needed regarding determining the potential importance of CD38 mediated NAD biology in MS pathogenesis. Here we review the functions of CD38 with special consideration devoted to its ability to control NAD levels.

Open in a separate window

Fig. (5)

Structures of a new ADP cyclase inhibitor and the first clinical trial inhibitor of constitutive HDACs, not Sirtuins are shown.

Knockout studies indicate that CD38 exerts dominant control of basal tissue NAD levels. Aksoy et al. propose that CD38 is the major enzyme negatively affecting tissue NAD levels [153]. Absence of CD38 results in high increases in steady state NAD levels. Elevations in lung and muscle are particularly dramatic with increases of approximately 10 [154] and 5 fold [155] respectively. Estimates of changes in brain tissue NAD levels range from approximately 2 [154] to 10 fold [153]. Other tissues possess as much as 20 fold greater levels of NAD [153]. Changes in NAD levels are typically fractional for most stress responses with the exception of PARP-1 activation by mutagen treatment, which can lead to decreases in NAD levels by over an order of magnitude [162]. However, increases in NAD levels of this magnitude are rarely if ever observed. It should also be realized that CD38 has long been used as a marker for hematopoietic stem cells, and has recently been determined to be constitutively present in the nucleus of a wide range of hematopoietic cells [163].

Where the first observations of the CD38 knockout mouse led to the understanding of CD38 as important for innate and adaptive immune function, the recent application of a high fat diet to the CD38 deficient mouse revealed profound roles for CD38 in controlling metabolic bioenergetics [155]. Barbosa et al. discovered that CD38 deficient mice fed a high fat diet do not gain weight. This was then determined to be due to increased mitochondrial biogenesis through the pathway extending: NAD-SIRT1-PGC1alpha-mitochondrial number. Accordingly, the direct application of NAD is now being considered in the context of obesity drug discovery research [164]. The metabolic implications of increasing NAD levels to allow the body to distribute NAD as needed may have tremendous therapeutic potential in the context of disease treatment.

As mentioned above, CD38 activity is constitutive where it is primarily regulated through changes in expression levels and perhaps sub-cellular location. Activation of CD38 by Th1-derived cytokines can cause decreases in available NAD. Immunomodulatory factors strongly regulate the expression level of CD38. Given the ability of CD38 strongly regulates basal NAD levels, we may predict that Th1-derived cytokines may cause a lowering of extracellular NAD levels. This may play a particularly important role in pathogenesis especially given that NAD has a much greater ability to protect neurons specifically in axotomy experimentation [27, 36]. TNFα-mediated changes in NAD levels appear to occur through altering CD38 activity more so than through altering IDO or other potential pathways [136, 165]. TNFα treatment causes a slight decrease in NAD levels of approximately 10% after 12 hours [136]. TNFα dramatically regulates CD38 expression levels more than any other NAD metabolizing enzyme (+100 fold within 6 hours), while NAMPT expression is elevated by approximately 10 fold. The pathways and mechanisms regulating cellular NAD supplies during inflammation are described in further detail below. Significantly, TNFα-mediated decreases in NAD levels are achieved through regulation of CD38 activity in macrophages [136]. Striking increases in CD38 expression have been seen in immunosuppressive B cells [166] and T cells [167]. Thus, CD38 in coordination with IDO is capable of mediating immunotoleragenic effects.

Another theory proposed is that CD38 may be functioning as a chemosensor for plasma NAD that has leaked from infected and damaged cells. Leucocytes respond by moving towards the NAD gradient. CD38 activity is required for proper lymphocytes chemotaxis. Similarly LPS stimulates CD38, which via the production of cADPR causes calcium release, which increases iNOS, and TNFα release from these activated microglia [139]. Inhibition of the CD38 activity prevents the entire process. It is known that approximately 100 NAD molecules are degraded per each individual cADPR molecule that is produced [153].

One of the first lines of defense preventing CNS disease is the blood-brain barrier (BBB). The brain is considered an anatomic site of immune privilege, whereby lymphocytes are not typically found in the brain. Disruption of lymphocyte trafficking across the blood brain barrier is considered as an attractive therapeutic goal [168, 169]. However, animal models of MS as well as in patients reveal the presence of high levels of lymphocytes in the brain. These cells are known to express the neuronal NAD-depleting enzymes IDO and CD38, the latter of which is required for microglial activation and dendritic cell migration.

Significantly, TNFα leads to decreases in NAD through increased expression of CD38 [136]. Microglial IDO is similarly regulated in response to LPS and TNFα, where IDO is known to decrease the concentration of extracellular NAD precursor tryptophan (for a review see [142]). The importance of CD38 mediated NAD depletion in microglial mediated pathogenesis is likely to be very significant indeed.

Anti-CD38 antibodies most of which activate calcium signaling have been observed in a significant number of type I and type II diabetics (for a review [170]). Similarly over-expression of CD38 compromises beta cell functioning. Collectively it would appear that auto-reactive CD38 antibodies might promote autoimmune disease. The importance of CD38 in MS pathogenesis is unexplored, but may be important in MS through Th1 cytokine-increased CD38-decreased NAD/increased calcium signaling-dependent pathways.

8.2. IDO Induction

Approaches to targeting IDO for increasing immunotolerance are extensively described elsewhere [142]. Thus, we only briefly describe a novel approach involving the first clinically available histone deacetylase (HDAC) inhibitor suberoylanilide hydroxamic acid (SAHA; Figs. 5, ​,6;6; reviewed [227]). SAHA was recently determined to increase IDO production, decrease cytokine production, decrease T cell proliferation, and to ultimately provide immunosuppression in the context of transplantation [228]. These effects were genetically determined to be dependent on increased IDO activity. SAHA is an inhibitor that works against the HDACs (class I and II) with constitutive activity, but not the dynamic NAD-dependent class III Sirtuins HDACs. SAHA is known to increase expression of neuronal SIRT1 [229]. Collectively, this suggests the potential existence of a HDAC rescuing SIRT loop whereby the cell responds to loss of constitutive HDAC activity by increasing SIRT1 expression, IDO activity, and therefore likely increasing NAD which may then increase SIRT1 activity (Fig. 6). This is a unique example connecting increased IDO activity with increased SIRT1. Ultimately, SAHA-mediated IDO-NAD-SIRT1 pathways may provide benefits in the context of MS.

Open in a separate window

Fig. (6)

Inhibition of the constitutive HDACs leads to increased transcriptional expression of SIRT1 and IDO. The increased IDO activity can increase NAD levels which can then further increase SIRT1 activity. Ultimately, SAHA increases immunotolerance. Thus SAHA may have potential in the context of controlling autoreactive pathogenesis in MS.

8.3. SIRT1 Activation

Activation of SIRT1 using resveratrol/SRT501 [104, 230], or nicotinamide riboside/SRT647 [104] has been shown to minimize the clinical signs and inflammatory response in EAE, while treatment with the SIRT1 inhibitor Sirtinol would prevent these beneficial effects [104]. Here we briefly review the current state of Sirtuin knowledge with emphasis on potential targeting in the context of MS. The Sirtuins are histone-deacetylating enzymes that exert global changes in chromatin structure in response to changes in either the NAD concentration or the redox status of the cell as measured by NAD/NADH ratio. Whereas most other deacetylases are considered to possess constitutive activity that is generally heritable, the Sirtuins are distinguished by their ability respond to metabolic fluctuations in NAD levels. One particularly exciting prospect from a pharmacological standpoint is the following. While we cannot change our genetics, we can change some of our epigenetics. Most epigenetic modifications are heritable and not alterable either. However, as the defining class III histone deacetylases, the Sirtuins can be targeted by pharmacologically modifying NAD levels. Functionally, the Sirtuins are extremely highly conserved evolution. They serve pivotal roles exerting control of metabolic functions in response to nutrient availability as observed in all examined organisms ranging from man to bacteria [231] !

In man there are seven different Sirtuins. SIRT1, SIRT6, and SIRT7 are primarily nuclear. SIRT3, SIRT4, and SIRT5 are primarily mitochondrial. SIRT2 is primarily cytosolic [232]. However, SIRT3 is also detectable in the nucleus and is the only Sirtuin family member clearly implicated in controlling human lifespan to date [233], while SIRT1 is also observed at lower levels in the cytoplasm [234]. As the founding member of the Sirtuin family, SIRT1 is by far the most studied isoform to date.

SIRT3 continues to emerge as a tremendously interesting Sirtuin family member. It is primarily expressed in the mitochondria where it functions in controlling both total cellular ATP levels [119] and global mitochondrial acetylation [120]. SIRT3 activity can protect against stress-induced cardiomyocyte cell death [233]. SIRT3 is predominately expressed in the mitochondria and has also been detected in the nucleus [233, 235, 236], however, other focused studies suggest SIRT3 is exclusively mitochondrial [232, 237]. In either case, by increasing SIRT3 expression in an NAD dependent fashion we expect to the level of ATP produced by the mitochondria to be increased based on previous studies [119]. The classic SIRT1 activator resveratrol is now known to also increase SIRT3 expression [121]. Accordingly it stands to reason that the ability of high dose nicotinamide to provide NAD and subsequently ATP may involve the NAD mediated activation of SIRT3 leading to sustained levels of ATP. Given that nicotinamide when applied at high concentrations is protective against Wallerian degeneration and EAE, greater attention should be directed at considering nicotinamide in the context of clinical MS.

Originally, the Sirtuins came into the limelight when yeast sir2 was identified as playing central roles in controlling caloric restriction (CR) mediated increases in lifespan [238]. CR represents 60% of normal caloric intake, where this is the only method that has consistently been shown to cause an increase in organismal lifespan. Accordingly large small molecule screens were performed to identify molecular activators of the human ortholog SIRT1. This lead to the realization that resveratrol is a potent activator of SIRT1 [239]. More recently SIRT1 function has been shown to be required for CR mediated increases in mammalian lifespan [16]. Further studies revealed SIRT1 deficient mice develop an autoimmune disease [240]. Since this realization, resveratrol has been verified for its ability to increase lifespan under normal caloric conditions in yeast, C. elegans, Drosophila, a teleost, and a mouse (for a review [241]). Resveratrol is a polyphenolic molecule produced by plants in response to pathogenic infection (a phytoallexin) that is abundant in grape skin and wine. Some have suggested that it is the primary reason explaining the French paradox [242, 243].

It should be mentioned that man actually empirically discovered resveratrol's benefits in the form of many plant-based medicines over 5,000 years ago [244]. Research devoted to increasing the molecular basis of resveratrol's mechanism of action has exploded since the discovery of its tremendous promise as a cancer therapeutic in particular [241]. With this increase in research it has become evident that there are many molecular targets of action for resveratrol application, where some neuronal growth stimulatory processes have been shown to be independent of SIRT1 activation in cell culture [245].

Resveratrol clearly activates SIRT1 [239] and SIRT2 [105], while increasing the levels of SIRT3 [121] - a master regulator of cellular ATP levels [119] and mitochondrial acetylation [120]. Thus, resveratrol may be a general positive regulator of Sirtuin activities.

Interestingly, SIRT2 is enriched in oligodendrocytes and its expression correlates directly with myelination based on analysis of proteolipid protein (PLP) deficient conditions [246]. Enriched in neurons, HDAC6 may seve complementary functions as a deacetylator of alpha tubulin [247]. While SIRT2 is generally cytosolic, it does in fact deacetylate histones during M phase to regulate mitosis [248]. Much more research is needed to elucidate the potential roles of NAD-dependent SIRT2 activity in controlling neural stem cell proliferation and remyelination.

However, several resveratrol-mediated phenomena are independent of SIRT1 and in fact the mechanisms of action for resveratrol are many (for an extensive book on resveratrol [244]). Resveratrol strongly promotes neurite outgrowth in Neuro2a cells in a SIRT1 independent but AMPK-dependent fashion [245]. AMPK can also be activated by calcium signaling pathways that are independent of AMP:ATP ratios and this appears to be the mechanism by which resveratrol works. Whether this may be through o-acetyl-ADP-ribose produced from SIRT1 activation or somehow ADP cyclase dependent processes, is unknown. Several polyphenols are also now known to activate AMPK [249]. This ability of resveratrol to promote neurite outgrowth may be of particular significance in the context of recovery from MS. Resveratrol also directly inhibits fatty acid synthase [250], promotes adipocyte apoptosis [121], and has been shown to promote fat loss in vertebrates [164]. Recently, it was determined that SIRT3 functions in the mitochondria as a major regulator of ATP levels in the mitochondria [119], while resveratrol is now known to cause increases in SIRT3 expression in adipocytes [121].

Recently, SIRT1 was determined to play a critical role in the fate specification decision for neural stem cells (NSCs; [251, 252]). Under reducing conditions, SIRT1 specificies neuronal differentiation, while under oxidative conditions SIRT1 specificies glial differentiation. One rationale for such decisions would likely be that the glia, as a supportive and biochemically superior cell, can then be used to deal with the oxidative stress which itself may be more likely to prevent neurogenesis. Accordingly, this presents interesting possibilities when one considers pathways from nicotinic acid. Application of high doses nicotinic acid is known to increases NADPH, and reduced glutathione, thus leading to more reducing conditions, while also decreasing angiotensin induced inflammatory R(O/N)S production and nuclear factor- κB (NF-κB) activity [253]. Further, applying NAD precursors increases NAD levels. Collectively, this may have the effect of providing a favorable reducing environment while simultaneously activating SIRT1 (Fig. 7). Thus this has the potential to favor neuronal regeneration. Application of niacin to the pellagric mental hospitals of the 1940s resulted in recoveries that were immediately obvious in many patients that had previously been listless, semi-catatonic, apathetic, and lacking in appetite [5]. The most common pathology observed in the brains of pellagrins involves neurodegeneration described as chromatolysis [147]. The potential for using nicotinic acid or other NAD precursors for stimulating neuronal regeneration is unclear.

Open in a separate window

Fig. (7)

Pharmacological nicotinic acid administration is predicted to favor neuronal genesis over gliogenesis due to the combination of SIRT1 activation combined and increased reducing conditions. SIRT1 controls redox mediated fate specification from neural stem cells. Pharmacological application of nicotinic acid increases NAD which activates SIRT1 (1) while also promoting reducing conditions by increased NADPH and GSH (2). Reducing conditions, being less stressful, allow neuronal specification instead of support cells. Otherwise under oxidizing conditions supporting glial cells are likely to be more immediately needed in order to correct the chemical imbalance, thus potentially favoring pathogenic gliosis. As discussed in section 2.1, minimizing gliosis can help foster remyelination. Thus the dual activity that nicotinic acid exerts by providing both NAD and reducing conditions is expected to respectively increase SIRT1 activity and the neuronal fat specification from stem cells that is likely to be needed for a complete and sustained recovery from MS attacks.

Another cellular strategy commonly considered for treating MS involves attempts to minimize activation of microglia [97]. SIRT1 is the focus of strategies to halt microglial activation as well through SIRT1 mediated activation of the transcription factor Liver X Receptor (LXR) or through inhibition of NF-κB as well (for a review [254]). Thus, SIRT1 activators persist as an increasingly area of investment for pharmacological drug discovery.

8.4. PARP-1 Inhibition

8.5. CD38 Inhibition with 2,2'-Dihydroxyazobenzene/Dyhydroxybenzene

Attempts are underway to consider CD38 inhibitors for their potential use in controlling inflammation or at least as molecular probes for basic research. Most recently 2,2'-dihydroxyazobenzene/2,2'hydroxybenzene (DHAB/DAB; Fig. 5) was realized to perform well as an inhibitor of the second phase calcium response in a rat model of cardiac hypertrophy (Cardiovascular Research Inhibition of ADP-ribosyl cyclase attenuates angiotensin II-induced cardiac hypertrophy Gul et al). In these experiments, angiotensin II (AII) treatment alone caused an initial release of calcium stores via IP3 signaling that was followed by a a sustained release of intracellular calcium via the CD38 catalyzed product, cADPR. Accordingly, inhibition of CD38 prevented the secondary sustained release of calcium, thus resulting in prevention of cardiac hypertrophy.

AII induced cardiomyocyte hypertrophy and cell death can be prevented by preserving NAD supplies through PARP-1 inhibition [111] or via increasing expression of NAD-dependent Sirtuins [112, 233]. Inhibition of CD38 would be expected to prevent depletion by CD38 itself as well as any other NAD depletion that may arise secondary to the sustained release of intracellular calcium stores through CD38 mediated generation of the second messengers cADPR, NAADP, and ADPR. However, NAD levels were not examined in this study. So it is unknown whether the desirable activities seen with the CD38 inhibitor DAB may in part be due to maintenance of NAD levels.

CD38 deficient mice challenged by an enforced high fat induced obesity model resistant weight gain through a pathway starting from dramatic increases in basal NAD levels that then activate SIRT1, which leads to increased mitochondrial biogenesis [155]. This strongly suggests that maintenance of NAD levels may be a major mechanism of action for DAB. Similarly, application of NAD itself has recently been proven to be effective at promoting fat loss in zebrafish larvae [164]. Therefore, one obvious application for DAB would be in the treatment of obesity. However, it is quite possible that DAB may also be most useful in the treatment of MS.

As described in greater detail above, the CD38 knockout mouse revealed that CD38 serves critical functions in both innate and acquired immune systems [152, 157]. Constitutively expressed in hematopoietic stem cells [163], CD38 is required for chemotaxis of professional APCs [152] and activation of microglia [138]. Thus, inhibition of CD38 with DAB may exert desirable effects in preventing inflammation and microglial activation. Since azo bonds are generally toxic, it will be especially necessary to examine dose dependence and potential effects on secondary infections arising from a weakened immune system.

Given that CD38 deficiency causes a persistent elevation of NAD, it will be most interesting to determine how this affects multiple sclerosis. As described in section 4, similar persistent maintenance of NAD levels seen in the Wld^S mouse actually exacerbates pathogenesis in TMEV-IDD models of MS while decreasing pathogenesis in EAE models of MS [98]. If the effect is persistent in a manner resembling the Wld^S effect, then we may see a similar delay of EAE mediated neurodegeneration but exacerbation in the TMEV-IDD model. If however, CD38 is required for the lymphocyte infiltration seen in TMEV-IDD, then under CD38 deficient TMEV-IDD conditions pathogenesis may be reduced. Finally, DAB should be examined in both of these contexts at a variety of concentrations. These kinds of experiments have the potential to dramatically help increase our understanding of the potential for developing CD38 inhibitors as a potential therapeutic approach.

9.2. Axon-Myelin Interaction

Axons and oligodendrocytes can profoundly affect each other's development and maintenance (Fig. 8b). In the case of axonal influences upon oligodendrocytes, axonal loss results in degeneration of the myelin sheath. In clinical and experimental spinal cord injury and the optic nerve during the developmental period, oligodendrocyte apoptosis follows along the transected degenerated axons [84, 292-294]. Interestingly, Lassmann et al. [295] described that some ultrastructural change in myelin degradation following optic nerve transection was similar to the one seen in patients with MS. Experimental optic nerve transection has significant effects on myelin protein mRNA expression in oligodendrocytes of optic nerve; a decrease in MBP and PLP mRNA expression and an increase in myelin-associated glycoprotein mRNA at 14 and 28 days after transaction [296].

Since axotomy also causes marked changes in the other glial population, including astrocytes, of the optic nerve, the alteration in myelin gene expression could ascribe to not only axonal loss but also astrocytosis and microgliosis. Nevertheless, this provides evidence that axons do influence myelin protein genes in oligodendrocytes. Similarly, stability of PLP mRNA has been shown to be decreased after axotomy in the PNS [297]. Although a decrease in myelin RNA has been demonstrated in TMEV infection (reviewed in [82]), damage in oligodendrocyte is thought to be a primary cause of down regulation of myelin RNA. However, as discussed above, myelin RNA can be down regulated secondarily to axonal damage.

Conversely, axonal preservation has been shown to require myelin proteins [289, 298]. Mouse mutants that lack expression of either PLP or 2',3'-cyclic nucleotide phosphodiesterase (CNP) have normal myelin sheaths, despite the fact that PLP and CNP are expressed exclusively in oligodendrocytes in the CNS [299]. However, aged PLP-null and CNP-null mice develop neurological signs, such as ataxia and hind limb paralysis, resulting from axonal degeneration. Human patients with a null mutation of the PLP1 gene develop axonopathy in the absence of major myelin abnormality. Interestingly, in PLP-null mice, the NAD-dependent deacetylase SIRT2, was virtually absent, while SIRT2 is expressed in oligodendrocytes in wild-type mice [246]. This suggests that SIRT2 may play a role in support of axons by oligodendrocytes, while SIRT2 has been suggested to play a critical modulating role in controlling morphological differentiation of oligodendrocyte precursors [300].

9.3. Microglia and Axon Interaction

Axonal (Wallerian) degeneration can induce local microglia activation, including up regulation of MHC class II molecules, without recruitment of peripheral lymphocytes (Fig. 8c) [85, 299, 301]. Block et al. [302] proposed that microglia activation and neuronal damage can lead to progressive neurodegeneration, regardless of the instigating stimuli. First, activated microglia initiates neuron damage by producing neurotoxic proinflammatory factors, including prostaglandin PGE₂, IL-1β, TNFα and superoxide. Second, microglia can become activated in response to neuronal damage (reactive microgliosis), which is toxic to neighboring neurons, resulting in a perpetuating cycle of neuron damage and microglia activation [302]. Interaction between CD200 on neurons and CD200R on microglia may play a role in this cycle.

When microglia activation and axonal degeneration occur in hosts who have encephalitogenic T cells in the periphery, this can lead to lymphocytes recruitment at the site of axonal degeneration in the CNS. In both EAE and TMEV infection, axonal degeneration with microglia activation can target inflammatory lesions at the site of Wallerian degeneration [290, 291].

Since the above three interactions can make a vicious cycle, intervention of any step can theoretically stop the cycle, leading to disease remission. However, many steps, including apoptosis, axonal degeneration, and microglia activation, have physiological beneficial roles. For example, axonal degeneration can prevent spread of neurotropic virus in the CNS, and activated microglia is important for innate immunity. As we discussed in this review, treatments that modulate NAD pathways can influence many different steps and cell types. Thus, the effects of such experimental treatments require careful assessment, which is necessary for clinical application of the treatment in future therapeutic trial in human patients.

We thank Daniel Doty, Andy Luu, and Faris Hasanovic for excellent technical assistance. This work was supported by the National Multiple Sclerosis Society (NMSS, PP1499) and the National Institutes of Health (NIH, R21NS059724). We are grateful to the critical comments and contributions of Christine Miller of Johns Hopkins University, Dharmesh D. Desai, Carmen Wheatley, and Soumya C. Chari for helping to prepare this manuscript. This provided critical firsthand insight into the mechanisms NAD mediated delay of neurodegeneration in EAE models of multiple sclerosis.