Introduction

The mitochondrial translocator protein (18 kDa; TSPO) was previously known as the peripheral-type benzodiazepine receptor (PBR) due to its ability to bind the benzodiazepine diazepam. However, its structure, expression, pharmacology and functions are different from those of the central benzodiazepine receptor [1] leading to renaming the protein TSPO [2]. The availability of high affinity specific TSPO drug ligands allowed for the assessment of the function of the protein. TSPO participates in many physiological processes, including metabolism and cellular respiration, cholesterol transport and steroidogenesis, immunomodulation, porphyrin transport and heme biosynthesis [2, 3]. It has also been proposed that TSPO may play roles in apoptosis and gluconeogenesis [4, 5].

TSPO is expressed in very low levels in the central nervous system (CNS) under physiological conditions; however, its expression levels increase in astrocytes and microglia as a result of brain injury and inflammatory processes [6], and thus it has been a subject of intense study interest, particularly during the past years. The binding of labeled TSPO ligands can be visualized and quantified by in vivo imaging techniques, and has become an important approach to study various neurological and psychiatric conditions. Despite TSPO has received a lot of attention in recent years as a target to image neuroinflammation and is the focus of this review, we will note that other inflammatory cells and molecules have been explored for imaging purposes as well in a diversity of neurological conditions involving inflammatory processes.

TSPO

Expression

The Tspo gene is evolutionarily conserved in most organisms [7]. In human beings, this gene is located on chromosome 22q13.3 and consists of 4 exons encoding a 169 amino acids protein [8, 9]. Although TSPO is expressed in many organs, its highest levels can be found in tissues containing steroid-synthesizing cells, such as adrenals, gonads and the brain [10]. Virtually all immune cells express TSPO [11]. In the brain, TSPO expression was considered to be specific for activated microglia and infiltrating macrophages; nevertheless, currently it is known that reactive astrocytes also express TSPO, although with a different spatiotemporal profile [5]. In addition, certain neuronal cell types have been also shown to express TSPO, such as those of the olfactory bulb [12] and dorsal root ganglia sensory neurons [13], as well as neural stem cells and post-mitotic neuronal precursors in developing or damaged brain regions [14].

Protein structure and binding

TSPO is a ubiquitous protein encoded by nuclear DNA, localized primarily in the mitochondrial outer membrane [15]; however, there is evidence that it may also localize to other subcellular structures such as the nucleus [16] and the plasma membrane [17].

TSPO’s protein structure, as initially suggested by its amino acid sequence, possesses five TM alpha helices with two extra- and two intra-mitochondrial loops. The five TM helices appear to be rigid while the terminal portions of the protein may remain flexible to allow the conformational changes necessary for protein–protein interactions. The N-terminus is located on the inside of the mitochondria while the C-terminus is located towards the cell cytoplasm and highly positively charged. A binding pocket is formed by the TM helices in the upper part of the extra-mitochondrial side, closed by a long loop between TM1 and TM2; nonetheless, additional binding sites may provide a higher level of the protein’s functional regulation [18–21].

It has been proposed that TSPO has a structure that facilitates cholesterol translocation [22–24]. A sequence denoted as the cholesterol recognition amino acid consensus (CRAC) motif is located at the C-terminal region of the protein, comprising residues 147–159 [23]. Within this motif, amino acids Y153 and R156 are believed to be critical for TSPO’s interaction with cholesterol [25].

Evidence exists that TSPO can polymerize by binding to other TSPO proteins or interacting with a range of different molecules. TSPO’s ability to form homopolymers appears to increase with mitochondrial activity and the generation of reactive-oxygen species (ROS) [26]. In support of these findings, recent structure–function studies demonstrated the role of TSPO in ROS generation [21]. Additionally, its protein structure is stabilized by ligand binding, which might mediate cholesterol transport by promoting binding to TSPO polymers [18]. However, polymerization has been found to increase ligand binding but reduce binding to cholesterol, suggesting an important involvement of polymerization in the mediation of TSPO function regarding cholesterol transport [20, 24, 26].

Specific mitochondrial proteins interact with TSPO, suggesting the presence of complexes formed by proteins from both the outer and inner mitochondrial membranes, including the 32 kDa voltage-dependent anion channel (VDAC) and the adenine nucleotide transporter (ANT) [27], together with other cytosolic and mitochondrial proteins [28]. These proteins include the mitochondrial permeability transition pore (MPTP) components, the peripheral benzodiazepine receptor-associated protein 1 (PRAX-1) [29], steroidogenic acute regulatory protein (STAR) and peripheral benzodiazepine receptor-associated protein (PAP7), a member of acyl coenzyme A (acyl-CoA) binding domain-containing proteins [28, 30] and ATPase family AAA domain-containing protein 3A (ATAD3A) [31]. Hence, it is likely that TSPO functions may be determined by the tissue- and cell-specific composition of mitochondrial membranes and mitochondria-associated organelles [32, 33]. The fact that cytosolic proteins can also interact with TSPO suggested a role of TSPO as a mitochondrial anchor transducing intracellular signals to mitochondria. As an example, it is suggested that acyl-CoA, or its binding proteins, may regulate TSPO function in the mitochondria and that TSPO participates in autocrine and paracrine signaling responses of glial cells to injury and pathogenic stimuli, mainly coming from the observations of TSPO interactions with endozepines in the peripheral and central nervous systems [3].

TSPO binds putative endogenous ligands, including cholesterol, protoporphyrin IX, phospholipase A2 (PLA2) and diazepam binding inhibitor (DBI) [34], a member of acyl-CoA binding domain-containing proteins [30], and a range of structurally diverse synthetic ligands including benzodiazepines, such as Ro5-4864 and diazepam, and isoquinoline carboxamide derivatives, such as PK 11195 [4]. In the CNS, PK 11195 reduces microglial activation and production of pro-inflammatory cytokines [35, 36]. How TSPO endogenous ligands, which are present at various levels in the tissues and cells examined, may affect endogenous drug ligand occupancy, affinity and residency time is unknown. This is a research question to be explored, considering the increasing use of radiolabeled TSPO drug ligands as imaging tracers (discussed later in this review) and potential changes in the levels of these endogenous ligands in various disease states.

Cholesterol and porphyrins show high affinities for TSPO, although porphyrins have affinity at the high nanomolar range, compared to the low nanomolar affinity for cholesterol [23, 37]. While cholesterol binds to the C-terminus, other ligands bind mostly to a region within the N-terminus [23, 38], although additional both steroidal and non-steroidal compounds binding at the CRAC motif have been recently reported [39, 40].

The classical/diagnostic synthetic ligands for TSPO are PK 11195 and Ro5-4864 (benzodiazepine 7-chloro-5-(4-chlorophenyl)-1,3-dihydro-1-methyl-2H-1,4-benzodiazepin-2-one). These ligands have been crucial for the characterization of TSPO’s expression and function and, particularly PK 11195, for the development of new TSPO ligands. For example, novel compounds synthesized to study TSPO’s binding properties have suggested the existence of multiple binding sites with possible allosteric effects in the human TSPO [41].

A single nucleotide polymorphism in the exon 4 of the TSPO gene, rs6971, resulting in the substitution of the amino acid alanine for threonine at position 147 of the TSPO protein (A147T variant), has proven to affect ligand-binding affinity [42, 43] of TSPO and affecting pregnenolone biosynthesis [20, 42, 44], although this might only be true for certain ligands as it was recently demonstrated that the variant shows an affinity to PK 11195 comparable to that observed for the wild-type protein, and retains the structure and dynamic profile [20, 45].

The ability of TSPO to bind cholesterol via the CRAC domain, and the ability of A147T polymorphism to affect cholesterol binding was recently confirmed in a series of structure–function studies where the presence of a cholesterol binding enhancement motif able to induce bacterial TSPO to bind cholesterol was also shown [19, 46].

The synthesis of the TSPO-specific ligand indol-acetamide FGIN-1-27 (2-[2-(4-fluorophenyl)-1H-indol-3-yl]-N,N-dihexylacetamide) [47] advanced the understanding of the TSPO pharmacology and led to the synthesis of a series of ligands which were safe in humans, such as emapuril (XBD-173; N-benzyl-N-ethyl-2-(7-methyl-8-oxo-2-phenyl-purin-9-yl)acetamide), which was shown in a clinical study to be safe and exert anti-anxiety activity while, in contrast to benzodiazepines, did not cause sedation and withdrawal symptoms [48].

Functions

It has been largely accepted that TSPO mediates various mitochondrial functions, including cholesterol transport and steroid hormone synthesis, porphyrin transport and heme biosynthesis, mitochondrial respiration, MPTP opening, calcium homeostasis, oxidation, apoptosis and cellular proliferation and differentiation [2, 49]. However, few of these functions have been directly demonstrated as most TSPO functions have been so far studied through its ligands’ actions. A function for TSPO in normal emotional regulation has also been suggested by the findings of a genetic association of the rs6971 polymorphism with bipolar disorder and adult separation anxiety disorder (ASAD) [50, 51].

TSPO can be found in intracellular locations other than mitochondria, such as the (peri)nuclear region and plasma membrane, playing different functions on a location-dependent manner. Nevertheless, non-mitochondrial TSPO, representing less than 5 % of TSPO [17], has received little attention so far. It is important to note that a Tspo paralogous gene, Tspo2, has been identified encoding an evolutionarily conserved family of proteins that arose by gene duplications [52]. Comparative analysis of Tspo1 and Tspo2 structure and function indicated that TSPO2 was characterized by the loss of diagnostic drug ligand-binding but retention of cholesterol-binding properties, and is involved in cholesterol redistribution during erythropoiesis [52].

The complex formed by mitochondrial TSPO in association with VDAC and ANT has been suggested to have a role in apoptosis, possibly through MPTP opening [34]. However, treatment with TSPO ligands has shown the ability to provide neuroprotection [53–55]. In fact, ligands such as PK 11195 and Ro5-4864 possess both pro- and anti-apoptotic properties, making PK 11195 and other TSPO ligands interesting targets for cancer therapies. Although the pro-apoptotic effects may or may not involve TSPO, the anti-apoptotic effects shown by these molecules are likely to take place through inhibition of TSPO’s apoptotic function, for which TSPO remains as a potential therapeutic target [11].

TSPO is thought to be important for tissue development and function. It may also participate in the biogenesis of mitochondrial membranes during cell proliferation and/or repair [3]. Furthermore, resulting from studies in animal models of neurodegenerative diseases, the observations that TSPO up-regulation in microglia and astrocytes associates with deleterious and beneficial effects, respectively [56], not only raise the possibility for a role of TSPO in regenerative processes but also suggest cell-specific functions. It has been shown that TSPO can modulate steroid production and, in turn, steroids are also able to affect TSPO’s ligand-binding properties [57]. A series of studies have supported a role for TSPO in inflammatory processes in peripheral tissues and the nervous system as a response to injury and disease, possibly through the regulation of steroid production [11, 58]. Nevertheless, recent studies have challenged the view of TSPO’s role in steroidogenesis, viability [59, 60] and MPTP [61], showing that knock-down/-out of TSPO in animal models does not affect steroid hormone biosynthesis, cell viability or MPTP activation even when induced by TSPO-binding molecules, suggesting that the mechanisms for these actions may not involve TSPO, as previously believed, and thus providing evidence to refute the major previously proposed functions for TSPO. Nonetheless, conditional knockout mice lacking TSPO in steroidogenic cells recently suggested that TSPO is indeed necessary for embryonic development during the pre-implantation phase and has a role in the mediation of stress responses [62]. Interestingly, similar discrepancy in genetic models has been reported for mitochondrial VDAC, where deletion of Vdac1 has been reported as both lethal and viable with minor phenotype [63]. Taken together these findings suggest that genetic models have to be analyzed with caution when dealing with evolutionary conserved proteins and conclusion should be reached when combining with biochemical, pharmacological and structural studies.

The TSPO-mediated pharmacology of cholesterol transport and steroidogenesis is well defined [24, 64], and supported by the biochemical and recent structural studies [2, 3, 19–21, 24, 45]. Indeed, TSPO drug ligands offered pharmacological means to regulate neurosteroid formation both in vitro and in vivo [3, 65]. This field expanded to neuropsychiatric and neurodegenerative disorders as well as neurotrauma [34, 66] and led to the use of TSPO drug ligands to alleviate neuropsychiatric disease symptoms mediated by increased neurosteroid formation in brain [3].

TSPO seems to be a sensitive biomarker of brain damage and neurodegeneration, particularly of inflammation and reactive gliosis. In the CNS, up-regulation of TSPO in response to damage is delayed in astrocytes as compared to microglia; however, the up-regulation in astrocytes is long-lasting, suggesting it may be crucial for its functions in neuronal survival and regeneration [67, 68]. Hence, TSPO expression can be expected to result modified in response to stressful stimuli and show alteration in diverse neurological and psychiatric conditions. A schematic representation of TSPO topology, effectors and functions is depicted in Fig. 1.

Fig. 1
figure 1

Schematic representation of TSPO topology, effectors and functions. TSPO is localized in the outer mitochondrial membrane, where it is found either alone or as part of a multiprotein complex together with VDAC and ATAD3. Cytosolic proteins, such as ACBD3, can also associate with TSPO. In active steroidogenic cells, this complex also contains the CYP11A1 enzyme responsible for the metabolism of cholesterol to pregnenolone, precursor of androgen, estrogen, mineralocorticoids, glucocorticoids and neurosteroids. TSPO drug ligands (e.g., PK 11195, Ro5-4864, FGIN-1-27, DAA1106, AC5216/XBD173, etifoxine), endogenous ligands (e.g., protoporhyrin IX, DBI and its metabolite TTN, PLA2), as well as associated proteins (e.g., ACBD3, PRAX1), could act as effectors and sometimes regulators of TSPO function. TSPO drug ligands could also serve as imaging agents to assess TSPO protein levels as related to various disease states (e.g., neurodegeneration, traumatic brain injury, cancer). TSPO functions, as assessed by the effects of its ligands, in mitochondrial respiration, cholesterol binding, import and steroidogenesis, porphyrin binding for heme biosynthesis, protein import for membrane biogenesis, control of free radical production, regulation of MPTP, apoptosis, cell proliferation, microglia activation and immune function

Full size image

Radiolabeled ligands and neuroimaging

TSPO is expressed at low levels in the normal brain but is locally up-regulated in sites of injury, possibly even before evident pathological and structural changes can be observed. This has provided a sensitive approach to accurately localize lesions and active disease processes [3] through the in vivo visualization and quantification of the binding of radiolabeled TSPO ligands as imaging agents for, mainly, positron emission tomography (PET) and, in a lesser extent, single-photon emission computed tomography (SPECT) and magnetic resonance imaging (MRI).

While Ro5-4864 and PK 11195 are the prototype diagnostic ligands for TSPO and have been long used to characterize the protein’s function, the development of new synthetic ligands has opened the doors for a wider array of applications which include, importantly, in vivo imaging of activated microglia and macrophage infiltration in the CNS, important markers of ongoing inflammation, in different pathological conditions. A range of synthetic TSPO ligands of diverse structural classes has emerged in past years, such as isoquinoline carboxamides, benzothiazepines and benzoxazepines, indoleacetamides, pyrazolopyrimidines, vinca alkaloids and aryloxyanilides, among others [69]. Examples of such compounds include DPA-713 and DPA-714 [70], DAA1106 [71] and its derivative, FEDAA1106 [72], PBR28 [73] and PBR111 [74], AC-5216 [75], CLINDE [76] and vinpocetine [77] (Table 1). Nevertheless, several characteristics need to be taken into account when evaluating these compounds, such as sensitivity, specificity, stability, clearance, species-specific metabolism and even the variable binding affinity in humans resulting from the rs6971 genetic polymorphism, for which new and improved TSPO ligands continue to be developed and evaluated for clinical purposes.

Table 1 Summary Table: targets and example agents to image neuroinflammation
Full size table

Some of the developed compounds may as well hold therapeutic potential for a number of neurological and psychiatric conditions, whether they selectively bind to TSPO or also bind to gamma-aminobutyric acid A (GABAA), or other types of receptors, which adds value to research in this field. For example, etifoxine, a benzoxazine, binds to TSPO and GABAA receptors and, while its anxiolytic effects have been suggested to involve the GABAA receptors [78], its neuroregenerative effects have been mainly attributed to TSPO [79].

TSPO neuroimaging applications

Up-regulation of TSPO expression in glial cells in response to injury and inflammation is associated with brain pathology [6] and its timing can track glial cell activation also during regenerative processes, which makes TSPO imaging a valuable tool to assess state, progression and repair in heterogeneous brain lesions, such as those resulting from traumatic brain injury (TBI) [66] and stroke [80]. Although up-regulation of TSPO has also been observed in the normal ageing brain using tracers such as [11C]PK11195 and [11C]vinpocetine [77, 81], suggesting that the activation of glial cells as well develops as part of the ageing process, the association between ageing and TSPO up-regulation remains controversial. A more recent PET study using the second-generation tracer [18F]FEPPA examined this association, finding no differences in TSPO expression related to normal ageing [82]. It is possible that the discrepancies between studies might be the result of not only the binding properties of different TSPO radioligands, but also of varying outcome measures and methods of analysis.

Up-regulation of TSPO at sites of neurodegeneration, and even more remote brain regions, has been observed in patients and animal models of diseases such as Alzheimer’s (AD), Parkinson’s (PD), Huntington’s (HD), amyotrophic lateral sclerosis (ALS), multiple sclerosis (MS) and frontotemporal dementia (FTD) [3, 83–85]. It is important to note that, in some cases (e.g., AD), the use of a specific TSPO tracer failed to confirm the increases in TSPO reported using other tracers [86]. In contrast to neurodegenerative diseases, and consistent with the role of neurosteroids as modulators of depression and anxiety [87], decreases in TSPO have been reported in a number of psychiatric disorders with anxious or depressive symptoms, including adult separation anxiety, post-traumatic stress disorder (PTSD) and schizophrenia [3, 88].

TSPO levels are increased in several types of cancer, including brain tumors, and thus it has been gaining attention in this area not only as an imaging agent but also as a target for the development of anti-cancer treatments [89–92]. In the peripheral nervous system, up-regulation of TSPO in Schwann cells, macrophages and neurons occurs in response to peripheral nerve injury, and TSPO ligands have shown to promote repair and provide neuroprotection [79, 93–96], in many cases likely to be mediated by increased neurosteroid production [3, 97].

Ten studies are currently registered using TSPO as a marker of neuroinflammation, mainly through PET imaging, in clinical trials for TBI, ALS, MS, PD, AD, mild cognitive impairment (MCI), schizophrenia, psychosis, major depressive disorder (MDD) and brain metastasis [ClinicalTrial.gov; accessed in June, 2015].

Other targets for imaging neuroinflammation

Besides TSPO, different types of molecules are also under investigation as targets to image ongoing neuroinflammatory processes in animal models and patients of a variety of neurological and psychiatric disorders using common and emerging imaging technologies (Table 1). One type of such is the cell adhesion molecules. Cell adhesion is essential for the migration of immune-competent cells to sites of injury, including leukocyte entry into the brain [98]. These molecules thus play an important role in inflammatory processes and have been targeted to observe infiltrating neutrophils, macrophages and T lymphocytes as well as activated platelets and endothelial cells using, for example, antibodies directed against the vascular cell adhesion molecule-1 (VCAM-1) and the intracellular adhesion molecule 1 (ICAM-1) conjugated with micron particles of iron oxide (MPIO) or paramagnetic liposomes [99–102], or a glyconanoparticle conjugated with syalil Lewisx (GNP-sLex) for the E- and P-selectins [103].

Strategies to measure oxidative stress-induced mitochondrial dysfunction and proteolytic activity have also been a subject of study to shed light on the inflammatory processes resulting from brain injury and disease. For example, approaches targeting free radicals through electron paramagnetic resonance imaging (EPRI), computerized electron spin resonance tomography (ESR-CT) or Overhauser magnetic resonance imaging (OMRI) using agents (or its precursors) with unpaired electrons, such as nitroxide radicals and methoxycarbonyl-PROXYL (MC-PROXYL) probes [104–107], have been developed. The myeloperoxidase (MPO) has been targeted using the MRI probe Gd-bis-5-HT-DTPA (MPO-Gd) [108] and luminol [109, 110]. Furthermore, a method has been proposed in which a combination of luminol and lucigenin bioluminescence enables the specific detection of acute (MPO-dependent) and chronic (NADPH oxidase-dependent) inflammation, respectively [111].

In response to CNS insult, infiltrating leukocytes, microglia and endothelial cells show increases in the expression of matrix metalloproteinases (MMPs) [112], whose activity is believed to play important roles in inflammation and blood–brain barrier breakdown. MMP activity has been visualized through non-invasive near-infrared fluorescence (NIRF) and, more recently, optical imaging, using activatable probes [113–115]. Imaging of MMP activity has been proposed as a useful measure to monitor anti-inflammatory effects [116] and, as well due to the important implication of MMPs in tumor formation, new combinatorial imaging methods are currently under exploration.

Much attention has been given to the cannabinoid receptor 2 (CB2), for which a diversity of PET tracers [117–124] and NIRF imaging probes [125, 126] have been developed. However, additional molecules implicated in a range of inflammatory processes have been targeted to image neuroinflammation, some examples include: cathepsin B, imaged using NIRF agents [127]; the cyclooxygenase 1 (COX-1), targeted using the PET tracer [11C]ketoprofen methyl ester [128]; P2X7, which has been gaining attention, targeted using the recently synthesized PET tracer [11C]A-740003 [129, 130]; I2-imidazoline receptors (I2Rs) detected using the PET tracer [11C]FTIMD [131]; and β-glucuronidase activity, visualized using [18F]FEAnGA for PET [132]. Toll-like receptors, receptor for advanced glycation end products, cytokines and chemokines [133] may represent good targets to image inflammatory status in the injured or diseased brain.

Finally, radiolabels such as technetium-99m (99mTc), indium-111 (111In), [18F]FDG and 64Cu enable the visualization of infiltrating leukocytes and may be used for labeling of specific leukocyte subpopulations. Phagocytic cells can be labeled using perfluorocarbons 19F and gadofluorine M. Another approach is to use iron oxide particles, such as ultra-small superparamagnetic iron oxide (USPIO), superparamagnetic iron oxide (SPIO) and MPIO, for in vitro or in vivo labeling [134].

Conclusions

Studies in animal models and early trials in humans suggest that the mitochondrial TSPO may be a sensitive biomarker of neuroinflammation and reactive gliosis. Although changes in TSPO expression are likely indicative of changes in mitochondrial function, the pathophysiological significance of increased TSPO expression in these processes is not well understood. However, the availability of specific imaging probes for TSPO makes this target attractive to assess the evolution and response to treatment of diseases with a major neuroinflammatory component.

In recent years, different types of molecules involved in inflammatory processes, other than TSPO, have been targeted for neuroimaging purposes. However, these studies have been relatively limited, thus rendering difficult the objective comparison of their advantages and disadvantages over TSPO imaging in the assessment of the inflammatory status of the injured CNS. Further studies on the most promising targets should shed light into their specificity, and the safety of the imaging molecules used to label them, compared to TSPO. Considering the complexity and dynamic nature of neuroinflammation, it is likely that more than one target may be required to assess its onset and progression.