Single-nuclei transcriptomic analysis of the subthalamic nucleus reveals different Pitx2-positive subpopulations

The subthalamic nucleus is a compartmentalized structure formed during development with each division serving a different function. A few molecular markers such as transcription factors involved in the formation and maintenance of the STN have been identified, and whilst a few more genes have been reported to be expressed in the STN, the complete transcriptomic landscape of the mature STN remains elusive. So far genes associated with the adult STN have been expressed throughout the structure. It therefore remains to establish if the functional division of this structure is attributed to a molecular diversity comprising of subpopulations. In this study a single-nuclei RNA-sequencing of genetically labelled Pitx2-Cre was conducted to identify the molecular heterogeneity of the STN. The findings revealed two subpopulations of the STN marked by the expression of Baiap3/Fxyd6 or Fgf11. Moreover additional Pitx2+ subpopulations with distinct signature markers that characterize neighboring structures of the STN such as the parasubthalamic nucleus, the lateral hypothalamus, the zona incerta and the mammillary nucleus were also identified.


Introduction
The subthalamic nucleus (STN) is a small excitatory structure located deep within the basal ganglia. The STN with its enriched glutamatergic projections plays a major role in motor, limbic and associative behavior (Temel et al., 2005) and is implicated in many neurological disorders, such as Parkinson's disease (PD), supranuclear palsy and Huntington, but also psychiatric disorders such as addiction and obsessive compulsive disorder (OCD) (Dickson et al., 2010;Lange et al., 1976;Limousin et al., 2005;Pelloux and Baunez, 2013). Some of these conditions are associated with degeneration of STN neurons, whilst others are correlated with alterations in its firing pattern within the basal ganglia circuity. The STN is a common target for electrical deep brain stimulation (DBS) aimed to alleviate motor symptoms PD and essential tremor and more recently STN-DBS has been introduced for addiction and OCD. The basic mechanism underlying the therapeutic effects remains unclear. It is also uncertain why in some cases DBS is accompanied by side-effects like apathy, depression and impulsivity (Temel et al., 2006). It has been proposed that the STN can be structurally subdivided into a tripartite structure in which motor, associate and limbic divisions form different connectivity and in turn mediate distinct functions (Hamani et al., 2004), raising the possibility that the side-effects reported by patients undergoing DBS may be linked to the position of the electrode.
Previous studies have proposed that the tripartite composition of the STN arises during development, in which the neurons migrate in a spatially coordinated manner from lateral to medial throughout three sequential waves (Altman and Bayer, 1979;Philips et al., 2005). The first set of neurons migrate from the ventricular zone to the dorsolateral part of the STN and form the compartment involved in movement, whilst the associative and limbic parts consist of neurons that migrate during the second and third wave of neurogenesis, respectively.
Transcription factors (TFs) are key regulators for determining the cell fate of neurons, as well their spatial and temporal acquisition, and may in turn contribute to the regional subdivision of the STN. To date a handful of TFs (FoxA1,FoxA2,FoxP1,FoxP2,Lmx1a,Lmx1b,Barhl1,Dbx1) have been shown to be expressed in subpopulations of the STN and associated with its development. However they are not exclusively linked to the STN, as some are also critical determinants of the midbrain (Gasser et al., 2016;Kee et al., 2016;Nouri and Awatramani, 2017;Skidmore et al., 2008;Zou et al., 2009). The gene encoding the TF paired-like homeodomain 2 (Pitx2) is expressed in the di-, mes-and rhombencephalon but serves as a key regulator for the STN. Mice lacking Pitx2 show a halted differentiation and migration of subthalamic neurons (Martin et al, 2004;Skidmore et al, 2008). Further, a Pitx2-Cre transgenic mouse line has been implemented to abrogate glutamatergic neurotransmission in the STN, which has resulted in identification of motor and reward-related features driven by the STN (Pupe et al., 2015;Schweizer et al., 2016Schweizer et al., , 2014.
To advance the understanding of molecular diversity within the STN and to identify molecular markers that could underpin the structural and functional preservation of the STN, single-nuclei sequencing approach of genetically labelled Pitx2-Cre population was employed to unravel the molecular landscape and heterogeneity of STN, with the scope of identifying whether the structural and functional division is a result of different neuronal subpopulations. Six Pitx2positive subpopulations were identified representing the MN, general hypothalamic populations, LHA/cSTN, PSTN, STN and ZI based on discriminate markers such as Foxb1, Fxyd6,Synpr,Cacna2d3,Col24a1 and Chrna4 that were upregulated in the individual clusters and their anatomical expression pattern evaluated from the Allen Brain Atlas. Importantly markers Baiap3 and Fxyd6 appeared to characterize one subpopulation of the STN cluster, whilst Fgf11 marked the second one. In addition compared to Pitx2, Stxbp2, Nxph1 and Nxph4 seemed to distinguish between the STN, PSTN, LHA/cSTN and MM with Stxbp2 expressed in all but the LHA cluster, Nxph1 in the STN and PSTN and Nxph4 in STN and MN. Finally, in this study a comprehensive gene list (Nmbr,Frmd4b,Dab2,Stard5,Pex5l,Galnt14,Trpc3) was found to be mostly expressed in the STN cluster and together with Htr2c, Nts, Nxph4, and Cdh23 can be used to study the STN in a more selective manner than Pitx2, which is expressed throughout the STN, but also in the neighboring structures.
Pitx2 has been shown to be expressed in the dorsal midbrain, the pre-and retromammilllary nuclei, various regions of the hypothalamus and the STN and to be a key regulator for the development of the STN but also the superior colliculus (Martin et al., 2004). However, whilst the development of the STN is most commonly thought to arise from the posterodorsal hypothalamic neuroepithelium or the closely related mammillary nucleus, some report the STN to develop from the ventrolateral thalamic neuroepithelium and consider the STN to be an intermediate zone of the ventral thalamus (Marchand, 1987;Philips et al., 2005). Interestingly, the different theories regarding the origin of the STN were also reflected in the results of the present study. Whilst most of the identified clusters are of hypothalamic nature or reside close to the hypothalamus such as the PSTN and the MN, the purple cluster that represents the ZI is marked by genes (eg. Kcnab3,Kcna1,Stard5,Bcan) that are shared between this and the red STN cluster, could potentially explain a more thalamic nature. Moreover, it seems that the t-SNE1 axis depicts the rostrocaudal migratory path with the red more anterior STN cluster and the MN cluster residing on the two opposite ends of the axis. This is also supported by the expression of Pitx2 and Calb1 and their co-localization detected in the MN. The t-SNE2 axis on the other hand probably reflects the dorsoventral or mediolateral pattern with the ZI and PSTN clusters in the two ends of the axis. mCherry Pitx2-Cre neurons were detected in the posterior and lateral hypothalamus, the MN, the PSTN and the STN, which is in accordance to the Allen Brain Atlas but also to previous studies addressing the expression of Pitx2 in development (Kee et al., 2016;Skidmore et al., 2008). However the current dataset provides further molecular markers that can distinguish between these brain regions beyond the TFs (eg. Lmx1a, Foxa1, Foxp1, Foxp2, Barhl1) identified in the aforementioned two studies. Of note, whilst the aforementioned TFs and Irx3 denoted the subthalamic lineage in the study by Kee et al. 2016, in the present study Irx3 was uniquely found in the PSTN cluster, although the possibility of sharing common progenitor lineage cannot be excluded. Moreover in this study the genes identified for the various clusters were associated with ion channels, transmembrane proteins, cell surface protein mediating, neuropeptides and neurotransmitter receptors among others with only the MN cluster showing a higher number of TFs. This finding is however not too surprising given that the mice used in this study were 28 days old, in which genes linked with cellular machinery rather than the differentiation process may be more essential.
Baiap3 and Fxyd6 were highly expressed in most clusters, similarly to previously broadly expressed markers Adcyap1 and Calb2 (Chen et al., 2017;Romanov et al., 2019). However the former were computationally distinctly expressed in only half of the red STN cluster. According to the Allen Brain Atlas, Baiap3 and Fxyd6 was expressed in the more medial part of the anterior STN. In the rodent this ventromedial part of the STN with its reciprocal connections to the ventral pallidum has been linked to associative and limbic functions (Tan et al., 2006). Given the anatomical expression of Baiap3 and Fxyd6, suggests that these genes may play a role in more limbic functions. So far Baiap3 has been shown to be significantly downregulated in the STN of mice exposed to 28 days of MPTP toxicity that could relate changes in STN firing with the transcript levels of Baiap3 (Lauridsen et al., 2011) but its role in behavior remains unknown.
Whilst it is still unclear whether the functional division with the different afferents and efferents of the STN are associated to gene expression, the current finding of the two subpopulations within the red STN cluster, marked by Baiap3 + /Fxyd6 + or Fgf11 + makes these genes prime candidates for further studies.
High frequency stimulation of the STN often results in adverse effects like depression which is believed to be linked with associated with serotonin depletion through inhibition of the midbrain serotonergic neurons (Temel et al., 2007). Interestingly, in the current study the second most upregulated gene of the red STN cluster compared to the rest is the serotonin receptor (Htr2c).
The rather high expression of Htr2c in the STN is in accordance to the Allen Brain Atlas and scRNA-seq dataset by (Wallace et al., 2017). As such this particular marker could be used further to explore the interplay of the STN and the serotonin system and possibly shed light in the adverse events reported with STN-DBS.
So far, little is known about the PSTN. It resides medially to the STN yet it is remains unclear how distinct it is from the STN and the LHA. One study reported the PSTN to influence hindbrain components of the central parasympathetic control network and to play a role feeding behavior and cardiovascular regulation, whilst differing from the LHA by a distinct population expressing β -preprotachykinin (Goto and Swanson, 2004). The findings obtained from the present study show that markers Cpne7, Hap1, Gpx3 and Fxyd7 that are highly expressed in the PSTN are also found in the LHA/cSTN cluster and general hypothalamic cluster, Stxbp2 and Nxph1 found in both STN and PSTN but not LHA/cSTN whilst markers like Ctxn1, Cacna2d3, Rasal1, Glra3 and Gck predominantly found in the PSTN cluster. As such all these markers in addition to Nxph4 that is found in the STN but not the other two clusters could be used in future studies to elucidate further the structural and functional role of the PSTN and possibly delineate the similarities or differences between the PSTN and the STN or LHA.         High Pitx2 High Fgf11 High both High Pitx2 High Baiap3 High both High Pitx2 High Fxy6 High both High Pitx2 High Stard5 High both High Pitx2 High both High Pitx2 High Stxbp2 High both High Htr2c High Stxbp2 High both High Pitx2 High Nxhp4 High both High Pitx2 High Nxph1 High both High Htr2c High Nxph4 High both High Htr2c High Cacna2d3 High both High Stard5 High Cacna2d3 High both (3) (4)  Nmbr Stxbp2