Positron emission tomography of type 2 cannabinoid receptors for detecting inflammation in the central nervous system

Abstract

Cannabinoid receptor CB₂ (CB₂R) is upregulated on activated microglia and astrocytes in the brain under inflammatory conditions and plays important roles in many neurological diseases, such as Alzheimer’s disease, amyotrophic lateral sclerosis, and ischemic stroke. The advent of positron emission tomography (PET) using CB₂R radiotracers has enabled the visualization of CB₂R distribution in vivo in animal models of central nervous system inflammation, however translation to humans has been less successful. Several novel CB₂R radiotracers have been developed and evaluated to quantify microglial activation. In this review, we summarize the recent preclinical and clinical imaging results of CB₂R PET tracers and discuss the prospects of CB₂R imaging using PET.

Cannabinoid receptor CB₂R

Cannabinoid receptors, including CB₁R and CB₂R subtypes, are G-protein-coupled receptors that are involved in endocannabinoid signaling and various physiological processes, such as pain, mood, and memory [1]. CB₁R is expressed abundantly in the central nervous system (CNS), whereas CB₂R exists mainly in the peripheral organs (e.g., spleen) and is undetectable in the brain under physiological conditions. In the periphery, CB₂R is expressed on leukocytes and macrophages and regulates inflammation and immune functions, such as phagocytosis inhibition, B and T cell differentiation, and the balance of pro-inflammatory/anti-inflammatory cytokines. Under CNS inflammation, CB₂R expression is upregulated on microglia, astrocytes, and oligodendrocytes and on post-synaptic neurons [2] in the hippocampus, brainstem, and cerebellum of both mouse/rat models [1, 3–5] and humans [6]. CB₂R has mainly been associated with anti-inflammatory/immuno-suppressive roles and reductions in nuclear factor NF-κB and inflammatory mediators. Its low baseline expression and disease-induced increase make CB₂R a promising diagnostic and therapeutic target for CNS inflammation.

Microglia and astrocytes are major players in the brain innate immune system. They are morphologically dynamic and undergo priming based on the local context [7]. Ramified microglia are activated in response to brain injury and immune stimuli and develop a “classical activation, pro-inflammation” status or an “alternative activation, anti-inflammation” status [8]. The extent to which inflammation is beneficial or detrimental in disease processes has not been fully elucidated. Acute inflammation consists of an immediate and early response to a stimulus, such as stroke or trauma, and a subsequent defensive response that facilitates repair. Chronic CNS inflammation resulting from persistent stimuli is an important component in the pathophysiology of many neurological diseases, such as Alzheimer’s disease (AD), Parkinson’s disease [9], amyotrophic lateral sclerosis (ALS) [10], multiple sclerosis (MS) [11], and Huntington disease [12].

CNS inflammation tracers

Positron emission tomography (PET) enables the in vivo detection of molecular changes in animal models and in humans, thus facilitating the understanding of disease mechanisms. The development of CNS inflammation tracers has been challenging, partly due to the diverse and changing states of astrocytes and microglia. Tracers for translocator protein (TSPO) have been the most applied probes in studying the microglial activation status in patients with neurological disorders [13, 14]. Several TSPO tracers have been evaluated in animal models and patients, including the first-generation [¹¹C]PK11195, second-generation molecules such as [¹¹C]DAA1106, [¹¹C]PBR28, and [¹⁸F]DPA-714 [15], and third-generation rs6971-insensitive radiotracers such as [¹¹C]ER176 [16] and [¹⁸F]GE180 [17, 18]. Although it has had varying degrees of utility in different neurological diseases, TSPO imaging has the limitations of specificity of binding, data quantification (reference region), and mixed binders of the second-generation tracers due to the polymorphism in the TSPO gene [14, 19]. New classes of tracers detecting, e.g., CB₂R, monoamine oxidase B [20] cyclooxygenases (COX 1, 2), and arachidonic acid [17, 21], are under development as potential tracers for CNS inflammation.

PET tracers for imaging CB₂R

Several CB₂R tracers have been developed, including the oxoquinoline derivatives [¹¹C]KD2 [22], [¹¹C]KP23 [23], [¹¹C]RS-016 [24, 25], [¹¹C]RSR-056 [26], [¹¹C]RS-028, and [¹⁸F]RS-126 [22, 25–31], and [¹¹C]NE40 [32]; and the thiazole derivatives [¹¹C]A-836339 [33, 34], [¹⁸F]JHU94620 [35, 36], and [¹⁸F]2f [37]. They have shown high-affinity binding to CB₂R and sufficient selectivity over CB₁R. The oxadiazole derivatives [¹¹C]MA2 and [¹⁸F]MA3 [38] and thiophene-based tracers [¹¹C]AAT-778, [¹¹C]AAT-015 [29], and [¹⁸F]FC0324 [39] show high affinity to CB₂R but suboptimal selectivity over CB₁R. Table 1 summarizes the relevant properties of the CB₂R radiotracers. Figure 1 shows the chemical structures of different classes of CB₂R tracers. For the structure-activity relationships of CB₂R tracers, see the recent review published by Spinelli et al. [40]. Some of the most important prerequisites for a useful CNS PET tracer for imaging CB₂R include: (1) high-nanomolar-affinity binding to CB₂R to obtain a high target signal in vivo (BP_nd or %ID/g) given the low concentration of CB₂R in the brain; (2) suitable LogP (range 1–3) for blood-brain barrier passage; (3) specific binding to CB₂R with 1000-fold higher selectivity over CB₁R; (4) no radiometabolites in the brain; and (5) low interaction with plasma proteins. This review focuses on the evaluations of CB₂R tracers in animal models and patients with neurological diseases involving CNS inflammation.

Table 1.

Summary of CB₂R tracer properties

Class	Tracer	Affinity hCB2R	Selectivity CB1R	LogP	Human	Animal model	Ref
Thiazole	[11C]A-836339	0.64 nM	>400	2.97		AD mouse model; stroke, LPS, AMPA injected rat model	[33, 34]
	[18F]2f	0.1 nM	>300	3.4		Healthy Wistar rat LPS-injected CD1 mouse	[41]
	[18F]JHU94620	0.38 nM	1000				[35, 36]
	[18F]15	0.29 nM	>1000				[37]
Oxoquinoline	[11C]NE40	9.6 nM	100	3.5	In vivo AD; healthy control	SD/Wistar rat stroke model; rhesus monkey	[32, 42–48]
	[11C]RSR-056	2.5 nM	>1000	1.94	Post-mortem AD brain, ALS spinal cord	LPS-injected mouse	[26]
	[11C]RS-016	0.7 nM	>1000	2.78			[25, 30]
	[11C]KD2						[22]
	[11C]KP23
	[18F]RS-126			1.99	ALS spinal cord		[23, 28]
	[11C]RS-028						[31]
	[11C]6f	5.4 nM	1851				[49]
	[11C]6b	9.4 nM	1067
Indole	[18F]FE-GW405833	27 nM	6000	2.5			[50]
	[11C]Methoxy-Sch225336	4.5 nM	78	2.15			[51]
Oxadiazole	[11C]MA2	87 nM	19			NMRI mouse	[38]
	[18F]MA3	0.8 nM	127
Thiophene	[11C]AAT-015	3.3 nM	303	4.11		Rat	[29]
	[11C]AAT-778	4.3 nM	256	6
	[18F]FC0324	0.1 nM	296			Rat	[39]
Amid	[18F]1,2	2.3 nM	>500			CD1 mouse	[52]
	[11C]10	11.5 nM	130				[53]

Fig. 1.

Chemical structures of CB₂R tracers of oxoquinoline, thiazole, oxadiazole, and thiophene-based derivatives

CB₂ imaging in animal models

Spleen tissue, which is rich in CB₂R, obtained from animals has been used to evaluate the specificity of CB₂R tracers. Several CB₂R tracers bind with high specificity in spleen tissue slices. Figure 2a, c show autoradiograms showing the specific binding using the PET tracers [¹¹C]RSR-056 and [¹⁸F]RS-126. The accumulation of [¹¹C]RSR-056 and [¹⁸F]RS-126 was blocked with the CB₂R antagonist GW405833 in spleen slices obtained from a CD1 mouse and a Wistar rat. Different animal models have been used for the in vivo evaluation of CB₂R tracers using PET imaging: (1) mouse and rat models with local human CB₂R overexpression, as well as rhesus monkey [42] and (2) lipopolysaccharide (LPS)-injected mice and rats. The uptake of CB₂R tracer in the brain is rather low, with a %ID/g of 0.1–0.4 at 1 h [33, 42] and a binding potential of 1.4–3 in animal models. Figure 3 shows the coronal view of [¹¹C]RSR-056-PET overlaid on structural magnetic resonance imaging, demonstrating higher radioactivity retention in the cortex and hippocampus of an LPS-injected mouse (b) compared to a vehicle-injected mouse (a) and an LPS-injected + GW405833-treated mouse (c). The PET images were generated from averaged dynamic data at 20–60 min after the intravenous injection of [¹¹C]RSR-056. The upregulation of CB₂R detected by using [¹¹C]RSR-056 appeared to be global instead of local, which might have been related to the specificity of the radiotracer. The results from a PET study using [¹¹C]A-836339 showed no specific uptake in the rat brain intrastriatally injected with LPS and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid [34], although immunostaining for CB₂R showed positive signals. Another PET study using [¹⁸F]RS-126 [28] or [¹⁸F]JHU94620 [35, 36] showed specific uptake in the CB₂R-rich spleen, but not sufficient signals in the brain, of LPS-treated mice. LPS-injected models have been used to represent a pro-inflammatory rather than anti-inflammatory state. Thus, they might not be the most suitable for evaluating of CB₂R.

Fig. 2.

Representative in vitro autoradiograms of CB₂R tracers. a [¹¹C]RSR-056 on spleen slices from a wild-type CD1 mouse. b [¹¹C]RSR-056 on cervical spinal cord from a healthy control and a patient with amyotrophic lateral sclerosis (ALS). c [¹⁸F]RS-126 on spleen slices from a Wistar rat. Left: baseline binding; right: blockade with CB₂R antagonist GW405833

Fig. 3.

Representative in vivo microPET images of coronal CD1 mouse brain sections (Bregma −1.5 mm ± 0.3 mm) by averaging dynamic data at 20–60 min after the intravenous injection of [¹¹C]RSR-056. a Vehicle-treated mouse. b LPS-treated mouse. c LPS-treated mouse with antagonist GW405833. Scale 0–0.5. SUV standard uptake value

CB₂R imaging in an animal model of stroke

Ischemic stroke involves a cascade of hemodynamic, vascular, structural, and inflammatory responses in a time-dependent manner [54–56]. Inflammation is implicated in the initial neuronal loss and consists mainly of pro-inflammatory responses and the extension of the lesion in the penumbra; the subsequent functional recovery involves anti-inflammatory responses. Upregulated CB₂R mRNA was reported in a mouse focal ischemia model, peaking at 5 days and returning to normal at 10 days after occlusion [54]. Treatment with CB₂R agonists has neuroprotective effects and attenuates macrophage/microglial activation in ischemic mouse models [54, 56–59].

The models of CB₂R imaging in ischemic stroke animal have been inconclusive. An increased [¹¹C]NE40 binding with no change in the binding of TSPO tracer [¹¹C]PK11195 [43] was reported in an SD rat model of photothrombotic stroke at 24 h after surgery. Another study using [¹¹C]NE40 imaging did not show an upregulated signal in a Wistar rat model of photothrombotic stroke [48], despite the minor increase in the level of CB₂R seen by immunostaining. Low [¹¹C]A-836339 uptake was detected in a rat model over 1–28 days after occlusion, despite positive immunostaining for CB₂R and CD11 (markers for microglial activation) [34]. The utility of animal models of cerebral ischemia has been demonstrated in previous studies using TSPO tracers. Possible reasons for the different observations include the time point of assessment and differences between the stroke models. Different methods for inducing acute stroke (transient or permanent ischemia) result in variations in size, the severity of the ischemic regions, and the expression of inflammatory markers [56]. Since the pathophysiology and inflammatory levels and types evolve during the course of ischemic stroke, it is critical to reduce variations and make assessments at the optimal times. As CB₂R are mostly associated with anti-inflammation, performing imaging, flow cytometry, and mRNA measurements at a subacute window 4–5 days after ischemia might be suitable [54].

CB₂R imaging in animal models of Alzheimer's disease

AD is the most common neurodegenerative disease. The abnormal accumulation of amyloid-beta (Aβ) and tau aggregates leads to a cascade of pathophysiological changes, including inflammation, microvascular alternations, synaptic dysfunction, and neuronal loss. Microglia play a phagocytic role in the clearance of pathological protein deposits [60] and act as pro-inflammatory agents [61, 62]. Treatment with CB₂R agonists in AD mouse models reverses Aβ-induced memory impairment [63]. MicroPET using [¹¹C]A-836339 showed specific uptake in the brain areas with Aβ depositions in an APPswe/PS1dE9 mouse model of AD [33]. PET studies in some other AD mouse models using [¹¹C]deprenyl in APPswe mice [20] and TSPO tracer [¹⁸F]GE180 in APP23 mice suggested increased astrocytosis and microgliosis at early disease stages. However, as age-related inflammation is also a confounding factor for the relatively mild inflammation in AD mouse models, these models might not be ideal for evaluating CB₂R tracers.

CB₂R imaging in animal models of multiple sclerosis

MS is an autoimmune inflammatory demyelinating disease [11]. In an experimental autoimmune encephalomyelitis (EAE) model of MS [64, 65], activated microglial cells had upregulated CB₂R at the protein (10-fold) and at mRNA levels (100-fold) compared to ramified microglia. Recent clinical trials on the CB₂R inhibitor delta(9)-tetrahydrocannabinol in patients with MS showed an ameliorating effect on the symptoms, including perceived spasticity and pain [66, 67]. Thus, PET tracers for CB₂R receptors have the potential to detect the pathophysiology of MS and monitor the effects of treatment targeting CB₂R in MS. However, in the EAE animal model, a clear disease status is considered difficult to define. Therefore, the EAE animal model might not be an ideal model for the evaluation of CB₂R tracers.

Potential clinical use of CB₂R PET imaging of CNS inflammation

Few clinical evaluations of CB₂R PET imaging have been reported [45, 47]. The hurdles in CB₂R tracer development and translation may include: (1) species differences between humans and rodents in their immune systems, microglia subtypes, and levels of CB₂R expression [68], as well as distinct age-related changes, with a more pro-inflammatory type in humans [69] according to transcriptomic analysis; (2) low expression level of CB₂R in the brain, even under inflammatory conditions [70]; (3) lack of specific CB₂R antibody for post-mortem validation of in vivo imaging results [43, 55, 71]; (4) lack of full CB₂R knock-out mice [72]; and 5) developing a reference region for modeling [73].

CB₂R imaging in patients with amyotrophic lateral sclerosis

ALS is a progressive and fatal motor neuron disease [74]. A variety of genetic factors drive the degeneration of motor neurons and increase the susceptibility to ALS or influence the disease progression rate. A greater density of CB₂R-positive microglia and macrophages was observed in the post-mortem spinal cord tissues from patients with ALS compared to healthy controls. Autoradiography using [¹¹C]RSR-056 showed higher and specific binding on the post-mortem spinal cord tissues from patients with ALS compared to that from healthy controls (Fig. 2b). Similar increases were reported in other studies on ASL spinal cord tissues using [¹¹C]RS-028 [31] and [¹¹C]KD2 [22].

CB₂R imaging in patients with Alzheimer's disease

Emerging evidence indicates the involvement of CB₂R and increased levels of complements, cytokines, and chemokines in the brain of patients with AD. Reactive astrocytes, activated microglia, and upregulated CB₂R were observed in the vicinity of Aβ plaques in post-mortem AD mouse model brains and patients with AD [62, 75–88]. PET using [¹¹C]NE40 showed reduced signal along with Aβ deposition visualized by [¹¹C]PIB in AD patients [47]. Further studies are needed to validate the utility of this tracer. A possible reason for these findings is that CB₂R upregulation is below the detection threshold, which is evident in the spinal cord from patients with ALS (Fig. 2b). In addition, as CNS inflammation is considered an early event, evaluating patients with mild cognitive impairment at the early stage of the disease process will be informative.

Outlook

We propose several points for consideration during the development of CB₂R tracers: (1) Subnanomolar-affinity CB₂R tracers should be developed, considering the low expression level of the receptor. (2) We should elucidate whether CB₂R upregulation is disease-stage dependent and relates to a certain type of microglia or astrocyte. (3) Structural optimization to improve the physicochemical and pharmacological properties of this new class of CB₂R tracer will be necessary. (4) Flow cytometry should be used for surface phenotype characterization, e.g., CD45 and CD11 positivity and differentiating pro- and anti-inflammatory microglial subtypes and astrocytes. (5) Imaging results using CB₂R and TSPO tracers should be compared head to head. Given that CB₂R more closely represents anti-inflammatory features, additional knowledge of the inflammatory status in the CNS will be provided by PET imaging using both CB₂R and TSPO tracers [43]. (6) The mRNA expression of CB₂R should be measured. (7) Rational translation evaluation processes will be vital. Mouse and rat models of cerebral ischemia at 4–5 days after reperfusion might be a suitable model to test and validate CB₂R tracers.

CB₂R is also expressed on activated immune cells outside the CNS which represent potential targets for CB₂R PET imaging. Recent CB₂R imaging using [¹¹C]RS-016 shows the detection of vascular inflammation in the post-mortem carotid plaque from humans and mouse models of atherosclerosis [30].

Conclusion

CB₂R PET ligands will serve as useful tools for tracking the in vivo alterations in CNS inflammation during the progression of these disorders and for understanding the disease mechanisms. CB₂R PET ligands will also help illuminate new treatment strategies for neurological disorders in which CNS inflammation is implicated.

Competing interests

The authors declare no conflicts of interest.