Cell-specific expression of trace amine receptor 1 (TAAR1) in neurons across brain regions: transcriptome analysis using publicly available databases

Relevance

Trace amine-associated receptor 1 (TAAR1) belongs to the G protein-coupled receptor (GPCR) family. It exhibits versatile signaling properties by coupling to multiple G protein subtypes, including Gs, Gq, and G13, thereby mediating diverse intracellular signaling cascades [1].

TAAR1 orchestrates signal transduction through a wide range of agonists, including endogenous trace amines, monoamine neurotransmitters, as well as exogenous psychostimulant drugs of the amphetamine class [2]. Trace amines are found in the mammalian brain at concentrations approximately 1000 times lower than catecholamines. It has been suggested that trace amines function as neuromodulators within classical neurotransmitter systems. TAAR1 has recently emerged as a promising therapeutic target for the treatment of neuropsychiatric disorders [3]. In particular, TAAR1 agonists are of interest as potential drugs for the treatment of schizophrenia due to their pronounced antipsychotic effects in rodent models and regulation of dopaminergic tone [4]. TAAR1 agonists also exhibit antidepressant/stress-reducing activity; pro-cognitive, eugeroic, antinociceptive, antinarcoleptic, and anticataleptic effects [5, 6].

TAAR1 agonists also include 3-iodothyronamine (3-T1AM) and thyronamine (T0AM), endogenous thyroid hormone derivatives that have been shown to cross the blood-brain barrier [7] and exert neuroprotective effects [8-10]. In particular, acting through TAAR1, thyronamines have been shown to inhibit the inflammatory response in astrocytes by modulating the release of pro-inflammatory (TNF-α, IL-6) and anti-inflammatory (IL-10) interleukins [11]. T1AM administration is capable of alleviating β-amyloid-induced neuronal dysfunction in wild-type mice [12]. In vivo studies have shown that 3-T1AM administration stimulates learning, completely reverses scopolamine-induced amnesia, and lowers the nociceptive threshold, causing hyperalgesia [13-15].

Despite growing interest in the physiological role of thyronamines, their precise mechanisms of action, toxicity profile, and impact on specific central nervous system cell populations remain insufficiently understood. One factor limiting progress in this field is the lack of detailed data on the precise anatomical and cellular localization of TAAR1 in the mammalian brain, particularly in humans [16]. Although TAAR1 expression at the mRNA level has been described for several brain structures, information regarding receptor distribution in specific neuron and glial cell types remains fragmentary.

Objective

To determine the expression pattern of TAAR1 in brain cells and tissues by analyzing publicly available gene expression databases.

The data on morphological brain structures with the highest TAAR1 density obtained in this work will be used to generate primary cell cultures enriched in thyronamine target cells, serving as a foundation for the next critical stage of research: studying the cytotoxicity and mechanisms of action of thyronamines. Furthermore, since TAAR1 represents a promising pharmacological target for neuroprotection and the therapy of various neurological and neuropsychiatric disorders, information on TAAR1 expression distribution may be useful for establishing a scientific basis for prioritizing this target in the development of innovative drugs within the domestic pharmaceutical industry.

Materials and Methods

Publicly available transcriptomic datasets of the brain providing comprehensive coverage of normal tissue and cell states of the human and rodent brain were used for TAAR1 expression analysis.

Relevant databases were searched using the Brain Knowledge Platform and FAIRsharing.org (query: brain gene expression).

Inclusion criteria:

• The database provides open access and clear documentation;

• Spatial mapping to brain anatomical regions and/or cell typing is available;

• Data pertain to expression in the human brain or model organisms relevant to neurobiological research;

• Data contain processed or normalized expression matrices.

Exclusion criteria:

• The database focuses on specific pathologies (e.g., Alzheimer's disease, schizophrenia, neurodegenerative diseases);

• The database covers only a single brain structure;

• TAAR1 is present in the final expression matrix, but expression levels are zero due to lack of detection, platform technical limitations, or library preparation features.

The data selection scheme included in the study is shown in Fig. 1.

Fig. Flow chart of dataset selection

The following datasets were included in the study:

1. Allen Institute Human Brain Atlas Microarray [17].
   Contains microarray data from 3,702 samples taken from various regions of six neurotypical adult human brains. Samples are distributed across cortical, subcortical, brainstem, and cerebellar regions in each brain and quantify expression levels of over 20,000 genes. Expression quantification is based on probe hybridization annotated according to the relevant human genome assembly versions (hg18/GRCh36) available at the time of platform creation. Data were processed using the abagen Python package version 3.14 (https://github.com/rmarkello/abagen), which employs preprocessing steps recommended by Arnatkevičiūte et al., 2019 [18].

2. Allen Brain Cell Atlas, "Neurons. Human brain cell type diversity" dataset [19].
   Includes human brain gene expression data obtained using snRNA-seq technology, aligned to the standard human reference assembly GRCh38, and organized using the t-SNE statistical method. In this dataset, all cells are classified into 31 superclusters — the broadest category, which hierarchically branches into 461 clusters and 3,313 subclusters. Each supercluster contains cells with similar general transcriptional or functional characteristics and is named based on the grouping of similar cells. Cells are divided across 13 anatomical divisions. Data analysis was performed using the interface available at https://knowledge.brain-map.org. The gene name was entered in the "Genes" tab, and anatomical structures and cell types were set using the "Neurotransmitter Type" and "Anatomical Division" filters. All cells and samples in which the target gene had a non-zero normalized expression value were included in the analysis.

3. BrainSpan Atlas of the Developing Human Brain (Allen Institute for Brain Science. Allen Developing Human Brain Atlas: Developmental Transcriptome) [20].
   RNA-seq data were obtained from tissues of 42 postmortem brains of physiologically healthy individuals of different ages, ranging from early prenatal development (8 post-conception weeks) to late adulthood (40 years). The brain was divided into 19 non-overlapping anatomical regions. RNA-seq data were annotated using gene and exon information from GENCODE version 10 (GRCh37–Ensembl 65). Exon expression levels were quantified in RPKM units (reads per kilobase per million mapped reads). The "RNA-Seq Gencode v10 summarized to genes" dataset was used for analysis. TAAR1 expression data were extracted using R version 4.5.2.

4. Whole Mouse Brain Transcriptomic Cell Type Atlas.
   This is a transcriptomic taxonomy of cell types across the entire mouse brain, integrating several single-cell RNA sequencing (scRNA-seq) datasets of the whole brain. The mouse transcriptome (M21, GRCm38.p6) was used as the reference transcriptome. The datasets contain a total of approximately 4 million cells. All cells are classified into 5,322 clusters organized hierarchically by class, subclass, supertype, and type [21]. Data analysis was performed using the interface available at https://knowledge.brain-map.org. All cells and samples in which TAAR1 had a non-zero normalized expression value were included in the analysis.

Results

1. Allen Institute Human Brain Atlas
   Before analyzing data from the Allen Institute Human Brain Atlas, we performed normalization and averaging across donors using the abagen tool. The initial steps of this process involve the following [22]:

• Microarray probes were re-annotated, and those not matching valid Entrez IDs were eliminated;

• Probes with levels below background noise and simultaneously in ≥50% of samples across donors were removed using intensity-based filtering.
  Already at these stages, all samples showing TAAR1 expression levels were excluded from further study. Thus, the microarray method employed in the Human Brain Atlas is insufficiently sensitive to assess TAAR1 expression levels. Therefore, datasets obtained using next-generation sequencing methods, including those allowing single-cell level sequencing, were selected for further work.

2. Allen Brain Cell Atlas
   In the "Neurons" dataset of the Allen Brain Cell Atlas, only 47 cells expressing TAAR1 were found. Expression levels ranged from 3.9 to 7.3 log2(CPM + 1). Thirty-three of the identified neurons belong to the Amygdala excitatory supercluster. These are excitatory glutamatergic neurons. Overall, among neurons expressing TAAR1, 40 were glutamatergic and 7 were GABAergic.

The distribution of neurons across brain structures is shown in Table 1.

Table 1. Distribution of neurons expressing TAAR1 according to the "Neurons" dataset by brain structures

3. BrainSpan Atlas of the Developing Human Brain
   In the BrainSpan Atlas of the Developing Human Brain data, TAAR1 expression was detected only in individual samples (Table 2). No expression was detected in any donor during the embryonic developmental period. The minimum age of a donor with detected TAAR1 brain expression was 4 months.

Table 2. TAAR1 expression levels detected in individual brain structures, according to the BrainSpan Atlas of the Developing Human Brain

4. Whole Mouse Brain Transcriptomic Cell Type Atlas
   In the "Whole Mouse Brain Transcriptomic Cell Type Atlas" dataset, 91 cells expressing TAAR1 were found (Table 3). Expression levels ranged from 3.1 to 7.1 log2(CPM + 1). Forty-seven neurons belonged to cluster 24 MY Glut. Regarding synapse type, 58 neurons were glutamatergic and 20 were GABAergic.

Table 3. Distribution of neurons expressing TAAR1 according to the "Whole Mouse Brain Transcriptomic Cell Type Atlas" dataset by brain structures

Discussion

The trace amine-associated receptor family was first identified in the work of Borowsky B. et al. [23] in 2001. The authors also attempted to describe the localization of TAAR receptors. Human TAAR1 gene mitochondrial RNA was detected using quantitative reverse transcription RT-PCR at low concentrations in individual regions of the central nervous system: the amygdala, cerebellum, dorsal root ganglia, hippocampus, hypothalamus, medulla oblongata, pituitary gland, and pontine reticular formation [23].

Widespread distribution of TAAR1 mRNA in the mouse CNS was revealed using in situ hybridization, and the hybridization signal was localized to the cytoplasm of neuronal profiles. Several brain regions showed intense labeling: the mitral cell layer of the olfactory bulb, piriform cortex, arcuate, motor, and mesencephalic trigeminal nuclei, lateral reticular and hypoglossal nuclei, cerebellar Purkinje cells, and ventral horns of the spinal cord. Moderate labeling was found in the frontal, entorhinal, and agranular cortex, ventral pallidum, thalamus, hippocampus, several hypothalamic nuclei, the ambiguous nucleus, dorsal raphe nucleus, and gigantocellular reticular nuclei. Weaker staining was observed in the septum, basal ganglia, amygdala, medulla oblongata, and dorsal horns of the spinal cord. Of particular interest was moderate TAAR1 mRNA expression in several monoaminergic cell groups, namely the dorsal raphe zone, locus coeruleus, and ventral tegmental area [23].

In the work of Bunzow J.R. et al., Northern blot of total RNA obtained from various brain regions, as well as in situ hybridization of rat brain sections, did not reveal the presence of TAAR1. However, using the more sensitive RT-PCR method, signals were detected from RNA obtained from various brain regions. Using this semi-quantitative method, it was determined that the signal appears to be widely distributed throughout the brain, with the highest expression levels detected in the olfactory bulb, olfactory tubercle, prefrontal cortex and other cortical areas, substantia nigra and ventral tegmental area, and some regions of the cerebellum, pons, and medulla oblongata [24].

To study the physiological role of TAAR1 in vivo, Lindemann L. et al. generated a mouse mutant in which the entire TAAR1 coding sequence was replaced by a reporter gene consisting of the LacZ gene, encoding bacterial β-galactosidase, and a nuclear localization signal (NLS) gene. In the targeted allele, LacZ is expressed using the endogenous TAAR1 promoter, providing a sensitive method for studying TAAR1 distribution in tissues [16]. RT-PCR experiments revealed low levels of TAAR1 expression in the mouse brain. Staining of serial brain sections from adult TAAR1 knockout and wild-type mice revealed discrete and specific localization of TAAR1-expressing nuclei, primarily in dopaminergic and serotonergic brain regions; in particular, the hypothalamus and preoptic area, ventral tegmental area, amygdala, dorsal raphe nucleus, nucleus of the solitary tract, parahippocampal region (rhinal cortex), and subiculum. TAAR1 was not detected in the olfactory bulb or Purkinje cells [16].

In the study by Revel F. et al., radioactive in situ hybridization failed to detect TAAR1 mRNA levels in the brains of wild-type mice. To further investigate the physiological role of TAAR1, the authors generated a genetically modified mouse line in which TAAR1 is overexpressed in neurons. However, this model does not allow assessment of TAAR1 expression distribution because expression is elevated in genetically modified mice, including in structures that do not normally express this receptor [25].

Thus, most studies indicate that, at least in the brain, TAAR1 is expressed at low levels. However, these data contradict the earlier report by Borowsky B. using non-radioactive in situ hybridization. Such discrepancies may result from species and/or methodological differences and illustrate the difficulties in reliably detecting Taar1 expression [25].

To study the role of TAAR in the development and maturation of dopaminergic neurons, TAAR expression was analyzed using TaqMan real-time PCR at various differentiation time points of three human induced pluripotent stem cell (iPSC) lines (AD3, WTSIi004-A, WTSIi032-A) differentiating into dopamine neurons. The authors found only sporadic expression of TAAR5 in the analyzed cell lines at all stages of the differentiation process and did not detect expression of any other TAAR types [26]. Analysis of the human HipSci iPSC cohort, which were differentiated into midbrain neurons, showed that TAAR1 expression could be detected at later stages of differentiation (by day 52) [26, 27].

Given advances in drug development targeting TAAR1, attempts have been made in recent years to image the receptor in the brain in vivo using positron emission tomography (PET). For example, the TAAR1 antagonist-based ligand [11C]TAAR1-1911 showed high binding affinity for the receptor in vitro but could not be used in vivo due to high rates of metabolic degradation [28]. The developed ligand [18F]TAAR1-2203 demonstrated high metabolic stability in vivo and allowed assessment of in vivo receptor occupancy using PET in the kidneys and pancreas. Signal intensity was markedly reduced in TAAR1 knockout animals. However, no specific binding to TAAR1 in the CNS was observed, although [18F]TAAR1-2203 appeared to cross the blood-brain barrier [2]. These results underscore the difficulties of imaging TAAR1 in the CNS under baseline conditions.

The results of our analysis of publicly available gene expression databases in brain cells and tissues confirm reports of low baseline TAAR1 expression levels in brain cells. Consequently, the microarray method does not allow assessment of TAAR1 expression because its level is below the noise threshold. More promising for research are next-generation sequencing methods, particularly single-cell and single-nucleus sequencing.

Comparison of the "Allen Brain Cell Atlas" and "BrainSpan" data shows partial overlap in the structures where TAAR1 expression was detected. These include various regions of the cerebral cortex, amygdala, hippocampus, and thalamus. Meanwhile, in the mouse brain, according to the "Whole Mouse Brain Transcriptomic Cell Type Atlas," the expression pattern differed significantly: expression was observed in cells of the medulla oblongata, midbrain, and hypothalamus. No expression was detected in cerebellar cells in any dataset.

TAAR1 expression was not detected at any embryonic developmental stage.

It is noteworthy that during transcriptomic classification, in both humans and mice, the majority of cells expressing TAAR1 belong to a single cluster. In both cases, these are glutamatergic neurons. Most neurons in which TAAR1 expression was found were glutamatergic by synapse type, followed by GABAergic neurons. Recent studies indicate that TAAR1 regulates glutamatergic transmission in a state-dependent manner — reducing spontaneous neuronal activity and enhancing electrically evoked glutamate release [3]. Although TAAR1 is thought to play a key regulatory role in modulating monoaminergic transmission, particularly in the prefrontal cortex, and there are reports that its mRNA localizes to monoaminergic nuclei [29], no such neurons were found in the Allen Brain Cell Atlas dataset.

Limitations

Substantial variability between data sources (differences in organism species, sequencing technologies, library preparation protocols) did not allow us to perform statistical analysis or meta-analysis. Therefore, we focused on a systematic review of publicly available expression data. Given these limitations, built-in normalization and counting methods available directly in the interfaces of the respective databases were used to obtain quantitative expression estimates. Mapping quality, read counts, and filtering criteria (minimum read number, quality metrics) were not recalculated.

Data on the proteomic expression of TAAR1 in brain tissues are of potentially significant interest; however, no databases meeting the established inclusion and exclusion criteria were identified during the search.

Conclusion

Overall, TAAR1 appears to be heterogeneously distributed in the central nervous system and is characterized by low baseline expression levels. Therefore, further research on its expression should employ state-of-the-art single-nucleus and single-cell sequencing technologies, as well as the development of in vivo receptor imaging methods. Analysis of publicly available gene expression databases in brain cells and tissues revealed the presence of TAAR1-expressing cells in human brain structures such as the cerebral cortex, amygdala, basal nuclei, hippocampus, and thalamus. Significant discrepancies in the observed expression patterns between human and mouse transcriptomic datasets suggest the possibility of species differences in TAAR1 expression.

Data Accessibility Statement

Materials and data from the Allen Institute for Brain Science may be freely used for research purposes with proper citation:

• Allen Institute for Brain Science (2010). Allen Human Brain Atlas: Microarray [Dataset]. Available from human.brain-map.org. RRID:SCR_007416

• Sten Linnarsson, Ed Lein, Trygve Bakken (2023) Transcriptomic Characterization of Cell Types in Human Brain. [Dataset]. Available from https://assets.nemoarchive.org/dat-5ie1mec

• Allen Institute for Brain Science (2010). Allen Developing Human Brain Atlas: Developmental Transcriptome [Dataset]. Available from brainspan.org. RRID:SCR_008083

• Hongkui Zeng, Zizhen Yao, Cindy van Velthoven, Kimberly A. Smith, Bosiljka Tasic, Changkyu Lee, Jeff Goldy, Anish Bhaswanth Chakka, Thuc Nghi Nguyen, Michael Tieu, Nick Dee, Junitta Guzman, Trangthanh Pham, Amy Torkelson, Rushil Chakrabarty, Tim Dolbeare, Nathan Guilford (2023). Whole Mouse Brain Transcriptomic Cell Type Atlas - 10x scRNAseq whole brain. [Dataset] Available from https://assets.nemoarchive.org/dat-qg7n1b0