This review starts with a background regarding the importance of monoamine oxidase B (MAO-B) function and role in neuropsychiatric illness; followed by development of PET imaging agents, focusing on those reaching human use. This is followed by discussion of radiotracers metabolized by MAO-B or irreversibly bound to MAO-B; vs. radioligands reversibly bound to MAO-B; and finally, some key clinical applications of MAO-B imaging. Background information will focus on the cellular and brain distribution of MAO-B, its functional role, abnormalities in postmortem tissue of neuropsychiatric diseases, and therapeutic inhibitors. Development of imaging agents will focus on [¹¹C]L-deprenyl analogs that use a radiotracer trapping approach for quantitation and [¹¹C]SL25.1188 which reversibly binds to MAO-B. Advantages and disadvantages of these approaches will be discussed. Clinical applications, mostly of quantitative methods, are then presented in major depressive disorder (MDD), cigarette smoking, and occupancy.

Monoamine oxidase has two isoforms in mammals, monoamine oxidase A and MAO-B, encoded by separate genes with about 70% common identity. MAO-B is an enzyme of 520 amino acids and is found in most tissues in the body including the brain (Saura et al., 1992, 1996). The structure of the human MAO-B enzyme includes an entrance cavity of 290A and a substrate cavity of 400A. In humans, MAO-B is found in dimers (Binda et al., 2011). In the brain, the predominant location for this enzyme is on the outer mitochondrial membranes of astrocytes and, within the midbrain, serotoninergic neurons, although there is also some MAO-B in dopamine containing cells within the substantia nigra (Damier et al., 1996; Saura et al., 1996; Arango et al., 2002; Lopez-Munoz and Alamo, 2009; Rajkowska and Stockmeier, 2013). In human brain, MAO-B is a high density target, abundant across many brain regions ranging from 1 to 5 ng/μg protein (and is two- to threefold higher than MAO-A) (Tong et al., 2013). MAO-B density is highest in the raphe nuclei in the brainstem (Saura et al., 1992, 1996; Tong et al., 2013). Available evidence indicates that MAO-B expression within these serotonergic neurons is largely confined to the nuclei, possibly because mitochondria with MAO-B are not transported to serotonergic nerve terminals (Arai et al., 2002). The serotonergic neurons of the raphe are responsible for most of the efferent serotonergic projections to the brain. Additionally, MAO-B density is very high in striatum and thalamus, high in the cortex and hippocampus, and relatively lower in cerebellar cortex and white matter tissue (Saura et al., 1992, 1996; Tong et al., 2013). Given the relatively lower expression of MAO-B in white matter, that fibrous astrocytes are mainly in white matter and that protoplasmic astrocytes are mostly found in gray matter, it could be posited that MAO-B is expressed more in protoplasmic astrocytes. However, it is also noteworthy that MAO-B has been identified in both protoplasmic and fibrous astrocytes (Levitt et al., 1982; Saura et al., 1996).

MAO-B has an important role in oxidation of amines including dopamine, norepinephrine, benzylamine, and phenylethylamine. The main catalytic reaction pathway involves oxygen reacting with the enzyme-product complex (see Figure 1). At the end of the pathway reduction of cofactor flavin adenosine dinucleotide reacts with O₂ to generate flavin and hydrogen peroxide (Binda et al., 2011). The metabolism of monoamines and production of hydrogen peroxide as well as aldehydes are of particular interest for understanding the relationship of MAO-B function to brain health. It is also proposed that MAO-B might facilitate production of toxic metabolic products as exemplified after MPTP administration. MPTP (1-methyl-4-phenyl-1,2,5,6-tetrahydropyridine) was originally a contaminant in the (illegal) synthesis of opiate desmethylprodine that induced Parkinson’s disease symptoms. Administration of MPTP in rodents results in behaviors of reduced and slowed movement which do not occur when MAO-B inhibitors are administered, implying that MAO-B metabolism of MPTP is required for the toxicity (Heikkila et al., 1984; Kupsch et al., 2001). However, it is noted that rats are not as susceptible to MPTP toxicity as mice (Lee et al., 1992). Generating oxidative stress adjacent to mitochondrial membranes and creation of potentially neurotoxic metabolic products has implicated a role for MAO-B in apoptosis and neurodegeneration (Riederer et al., 2004; Youdim et al., 2006; Mallajosyula et al., 2008).

The effects of overexpression of MAO-B and MAO-B knockout have also been evaluated. In a transgenic mouse model of globally increased MAO-B in astrocytes, behavioral changes during open field assessment included reduced total movement, less distance traveled, lower movement speed, and lower duration of time moving, effects attributed to midbrain neurodegeneration but it is noteworthy that the MAO-B overexpression was global across astrocytes throughout the brain (Mallajosyula et al., 2008; Siddiqui et al., 2011). The astrocytosis associated with MAO-B overexpression in this transgenic model was prevented by accompanying MAO-B inhibitor treatment (Mallajosyula et al., 2008; Siddiqui et al., 2011). In a knockout model of MAO-B, in a paradigm of repeated open field exposure designed to assess habituation to novelty, reduced habituation was observed, suggesting that greater movement and novelty exploration are associated with reduced MAO-B level (Lee et al., 2004).

MAO-B density and activity are of considerable interest as indices of astrogliosis, and as abnormalities in several brain illnesses. Greater expression of MAO-B has been observed in astrocytes in many neuropsychiatric diseases such as Alzheimer’s disease and amyotrophic lateral sclerosis (Ekblom et al., 1993; Saura et al., 1994). This may account for the high number of neuropsychiatric diseases associated with elevated MAO-B density as well as therapeutics developed (see Table 1). However, in regard to therapeutics, it should be acknowledged that the MAO inhibitor discovery and development for MDD was originally serendipitous and based on observations that patients with tuberculosis receiving treatments with non-selective MAO inhibitor properties then had reductions in depressive symptoms. Interestingly, indices of the concentration of glial fibrillary acidic protein (GFAP), a sensitive and reliable marker of most, if not all, reactive astrocytes during brain disease, demonstrate high correlations with MAO-B binding in the primary regions assessed across postmortem investigations in Alzheimer’s disease, multisystem atrophy, progressive supranuclear palsy, and amyotrophic lateral sclerosis (Ekblom et al., 1993; Saura et al., 1994; Tong et al., 2017). Interestingly, assessments of MAO-B level in relation to other markers of astrocytosis such as vimentin and Hsp27 in the putamen of multisystem atrophy and progressive supranuclear palsy have also shown high correlations (Tong et al., 2017).

An important consideration is whether to develop a MAO-B PET imaging agent as an index of MAO-B density with irreversible or reversible binding or aim to develop a metabolic substrate of MAO-B to reflect an index of MAO-B activity. The density of MAO-B is highly correlated with its activity; with correlation coefficients typically > 0.9; plausibly due to minimal variation in two post-translational modifications to MAO-B (Saura et al., 1992, 1994, 1996; Tong et al., 2013). Hence measures of density are highly functionally relevant. Development of PET imaging agents for MAO-B has approached several of these directions.

Development of MAO-B PET imaging agents started with the strategy of applying radiotracers irreversibly bound to MAO-B so as to evaluate an index related to MAO-B density, as exemplified by [¹¹C]L-deprenyl, [¹¹C]L-deprenyl-D2, and [¹⁸F]fluorodeprenyl (also see Table 2). The advantage of this approach is that the density of the enzyme is largely related to the concentration of trapped radiotracer. A disadvantage is that excessive trapping results in irreversible time activity curves, which makes separating of parameters corresponding to radiotracer interactions with MAO-B from parameters corresponding to radiotracer delivery (and blood flow) difficult. The optimal window of radiotracer trapping vs. maintaining adequately reversible time activity curves to better quantify parameters of specific binding is difficult to achieve. [¹¹C]L-deprenyl was the first PET MAO-B imaging agent applied in human brain in vivo (Fowler et al., 1987). Then to generate more reversible time activity curves, analogs were subsequently synthesized like deuterium labeled [¹¹C]L-deprenyl ([¹¹C]L-deprenyl-D2) (see Figure 2). [¹⁸F]deprenyl-D2 was also created and modeled in monkeys (Nag et al., 2016). The substitution of D2 reduced the rate of radiotracer binding by MAO-B leading to more reversible time activity curves. Another approach to obtain more information regarding specific binding parameters is to have a longer scanning period which was attempted with fluorination of deprenyl (with radiotracer [¹⁸F]fluorodeprenyl-D2; Nag et al., 2016). A limitation of this direction of radiotracer development is that [¹¹C]L-deprenyl (as well as [¹¹C]L-deprenyl-D2, etc.) are metabolized to brain penetrant radioactive metabolites which have additionally specific binding to dopamine and other monoamine transporters such as [¹¹C]methamphetamine and [¹¹C]amphetamine (Fowler et al., 1995). This complicates quantitation, especially in regions with high density of dopamine transporters like the striatum and substantia nigra.

A radioligand with an entirely different structure, [¹¹C]SL25.1188 (see Figure 2), was discovered by Bramoulle et al. (2008) and subsequently modeled in baboons (Saba et al., 2010). This radioligand has highly reversible time activity curves giving considerable advantage for identifying parameters related to specific binding. The first production method for [¹¹C]SL25.1188 required the toxic nerve gas phosgene (Bramoulle et al., 2008) making routine synthesis difficult. A new synthesis method involving CO₂ fixation was discovered by Vasdev et al. (2011a,b), and then the radioligand was modeled in humans. [¹¹C]SL25.1188 has outstanding properties for quantitative PET imaging, including high reversibility, brain uptake, and selectivity for MAO-B; no brain penetrant metabolites; and a very reproducible total V_T measure, an index of MAO-B level, which is highly correlated with the known concentration of MAO-B in postmortem human brain (r² > 0.9) (see Figure 3 and Table 2; Rusjan et al., 2014).

A fluorinated analog of this compound, [¹⁸F] (S)-3-(6-(3-fluoropropoxy)benzo[d] isoxazol-3-yl)-5-(methoxymethyl)oxaz olidin-2-one, [¹⁸F] FSL25.1188 has also been synthesized, with high affinity, high brain uptake, and excellent reversibility in cynomolgus monkey and high selectivity in gray matter of the brain. Blocking studies suggested a lesser overall reduction of radioligand uptake in white matter leaving the possibility of either different free and non-specific binding in this tissue, or non-selectivity. Metabolites are polar, suggesting they are unlikely to be brain penetrant. However progressively increased radioactivity in bone after radioligand administration suggests that there is potential for bias from radioactivity in this structure for brain regions proximal to the skull. Overall, this radioligand has excellent potential for human study for brain regions not near to the skull (Dahl et al., 2019).

There is also continued interest in developing radiotracers that are metabolized such that the radioactive metabolite is trapped in the central nervous system. The advantage of this approach is that the function of the enzyme is largely related to the concentration of trapped radioactive metabolites. A disadvantage is that MAO-B is found throughout bodily organs, with the exception of the placenta, so the same metabolism occurs throughout the body, so the radioactive metabolites would likely be found in blood plasma as well as brain. This raises the question as to what extent the radioactive metabolites found in the brain are related to metabolism of the parent compound in the brain vs. radioactive metabolites in plasma crossing into the brain. There are exceptions to this concern, for example, if the radiotracer is trapped within the tissue and not distributed to plasma. This is exemplified with the radiotracer [¹¹C]Cou for which the polar metabolite is trapped in tissue (Drake et al., 2018).

PET imaging agents metabolized by MAO-B with trapped radioactive metabolites is still an ongoing direction of development. This direction still shows intriguing promise given findings with the first radiotracer developed with PET, [¹¹C]MPTP. [¹¹C]MPTP had several useful properties, best demonstrated in its results in rhesus monkey showing high SUV; differentiation in uptake across regions consistent with regional variation in MAO-B density; and time activity curves that peaked reasonably early despite using a trapping mechanism to retain radioactive metabolites in the brain. However, this radiotracer was not intended for human use due to the severe and largely irreversible toxic effects associated with pharmacological doses of MPTP. The MPTP analog 4-methyl-7-((1-[¹¹C]methyl-1,2,3,6-tetrahydropyridin-4-yl)oxy)-2H-chromen-2-one ([¹¹C]Cou) was assessed in rhesus monkeys and found to be selectively metabolized by MAO-B; with the product of MAO-B oxidation and subsequent hydrolysis trapped in the brain (Drake et al., 2018). However, excessive trapping led to time activity curves that were insensitive to the relative density of MAO-B in the brain. Two deuterium substituted analogs, ([¹¹C]Cou-d3 and ([¹¹C]Cou-d7) were created with the intention of reducing MAO-B metabolism further, by slowing down the hydrogen abstraction step, implicated as rate limiting in the MAO-B oxidation of MPTP, but the impact on time activity curves was fairly minimal, suggesting that these two analogs should not be developed further (Drake et al., 2018).

There is also continued interest in developing reversible MAO-B PET radioligands that are fluorinated to have longer half-lives so that the radioligands could be delivered to PET imaging centers without onsite cyclotrons. [¹⁸F]THK-5351 was designed to detect neurofibrillary tangles in vivo and was later found to bind MAO-B with high affinity (Villemagne et al., 2018). Unfortunately, the non-selective binding of [¹⁸F]THK-5351 to MAO-B and tau limits its utility for either purpose. However, a similar structured compound, (S)-(2-methylpyrid-5-yl)-6-[(3-18F-fluoro-2-hydroxy)propoxy]quinoline ([¹⁸F]-SMBT-1) (Figure 2) has been synthesized and applied in humans, although to date published validation studies are lacking (Harada et al., 2021; see Table 2). Selectivity of this latter compound for MAO-B was demonstrated in blocking studies in autoradiography of Alzheimer disease brains (which were positive for tau based on higher uptake of [³H]-MK6240; Harada et al., 2021).

It is noteworthy that there are evident problems for using reference tissue methods for MAO-B PET imaging agents. The distribution of MAO-B in the human brain is generally high ranging from 1 to 5 ng/μg protein (Tong et al., 2013). Relatively lower density regions include the cerebellum and white matter tissue, but MAO-B is still high in these tissues (Saura et al., 1996; Tong et al., 2013). Even for these regions, the MAO-B concentration is approximately half to a third of the concentration in the cortex and striatum, respectively, unfavorable ratios for use of a reference tissue (Saura et al., 1996; Tong et al., 2013). Accordingly, valid reference regions seem unlikely. If such regions were used as a reference like tissues for investigation into disease, it would also need to be assumed that there were no disease effects in the structure. Also, the MAO-B density is likely to contribute to a separate compartment in a reference tissue so it would seem plausible that reference tissue models accepting of more than one compartment in the reference tissue would likely be preferred as compared to single tissue compartment reference tissue approaches. Given these issues, clinical research presented largely excluded studies applying reference tissue methods.

MAO-B is an important brain protein given its high expression in the central nervous system; and its role in generating oxidative stress, facilitating monoamine metabolism, and influencing astrogliosis. Abnormalities of MAO-B density and/or activity are found in postmortem study of many diseases including Alzheimer’s disease (Saura et al., 1994), Parkinson’s disease (Tong et al., 2017), progressive supranuclear palsy (Tong et al., 2017), multisystem atrophy (Tong et al., 2017), Huntington’s disease (Richards et al., 2011), amyotrophic lateral sclerosis (Ekblom et al., 1993), and alcohol use disorder (Udemgba et al., 2014).

PET imaging developments for MAO-B largely followed strategies of irreversible and reversible binding compounds. [¹¹C]L-deprenyl was the first radiotracer applied in human brain imaging of MAO-B with PET (Fowler et al., 1987), but to reduce the impact of poor reversibility, [¹¹C]L-deprenyl-D2 was created. [¹¹C]L-deprenyl-D2 has limitations of poor reversibility, and radioactive metabolites with specific binding for dopamine and norepinephrine transporters. It is particularly difficult to make imaging agents with a trapping compartment for the parent compound or metabolites and, at the same time, both maintain reversible time activity curves in brain tissue and avoid similar metabolites being generated in the periphery, as MAO-B is found through the body. However, this does not preclude the possibility of accomplishing this, and efforts continue with radiotracers such as analogs of [¹¹C]Cou. In contrast, [¹¹C]SL25.1188 is an outstanding radioligand with a specific binding to non-displacement ratio of 7–12, high selectivity, high reversibility, and no evidence for substantial brain penetrant metabolites. Its main disadvantage is that just like [¹¹C]L-deprenyl-D2, quantitation presently requires arterial blood sampling.

Important clinical research applications have been completed with MAO-B imaging. The elevation in MAO-B V_T of over 1.5 standard deviations in the prefrontal cortex during MDE of MDD is quantitatively substantive, and, in combination with the roles for MAO-B in generating oxidative stress, metabolizing monoamines, and apoptosis, makes a case that the MAO-B inhibition of MAO inhibitors, not just the MAO-A inhibition has important therapeutic implications. Regardless of the limitations of [¹¹C]L-deprenyl-D2, the interpretation of a lesser level of MAO-B binding in those who smoke cigarettes is solid and replicable. The mechanism underlying this interesting finding is uncertain although irreversible binding by a metabolite of cigarette smoke is a plausible mechanism. It was demonstrated that gingko biloba does not occupy MAO-B, and that sembragiline occupies a high proportion of MAO-B sites, whereas the occupancy study of rasagaline did not use a validated method of quantitation. Hence the main implication of this latter study is that rasagiline shows brain penetration of drug that persists over a longer time course than its removal from plasma, with the numerical level of occupancy being uncertain. While several important clinical applications have been completed with MAO-B imaging, given the number of neuropsychiatric diseases with abnormal MAO-B density and/or activity, and the role of MAO-B in astrogliosis, it is evident there are opportunities for more discovery and drug development.

JM: conception and primary draft manuscript preparation. JB: draft manuscript preparation. Both authors contributed to the article and approved the submitted version.

JM has received operating grant funding from the Government of Ontario with Janssen for studies of translocator protein binding in relation to blood biomarkers. JM also has patents for blood biomarkers in mood disorders to predict neuroinflammation and elevated MAO-B in the brain and a patent for a dietary supplement to prevent post-partum depression. JM was also developing MAO-B inhibitors for use in neuropsychiatry. The remaining author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.