The transcription factor BCL11A defines a distinctive subset of dopamine neurons in the developing and adult midbrain

Midbrain dopaminergic (mDA) neurons are diverse in their projection targets, impact on behavior and susceptibility to neurodegeneration. Little is known about the molecular mechanisms that establish this diversity in mDA neurons during development. We find that the transcription factor Bcl11a defines a subset of mDA neurons in the developing and adult murine brain. By combining intersectional labeling and viral-mediated tracing we show that Bcl11a-expressing mDA neurons form a highly specific subcircuit within the dopaminergic system. We demonstrate that Bcl11a-expressing mDA neurons in the substantia nigra (SN) are particularly vulnerable to neurodegeneration in an α-synuclein overexpression model of Parkinson’s disease. Inactivation of Bcl11a in developing mDA neurons results in anatomical changes, deficits in motor learning and a dramatic increase in the susceptibility to α-synuclein-induced degeneration in SN-mDA neurons. In summary, we identify an mDA subpopulation with highly distinctive characteristics defined by the expression of the transcription factor Bcl11a already during development.


1
Midbrain dopaminergic neurons (mDA) are anatomically organized into the substantia nigra, 2 ventral tegmental area (VTA) and retro-rubral field (RRF). These anatomically defined areas 3 contain subpopulations of mDA neurons that are characterized by distinct molecular profiles, 4 distinct connectivity and distinct impacts on dopamine-modulated behavior (Engelhard et al., 5 2019;Poulin et al., 2018;Poulin et al., 2020). The SN consists of the pars compacta (SNc), 6 pars lateralis (SNl) and pars reticularis (SNr). The majority of SN-mDA neurons is located in 7 the SNc, a smaller population forms the SNl and only few mDA neurons are found in the SNr.

8
SNl mDA neurons project to the tail of the striatum (TS) and have been shown to reinforce the 9 avoidance of threatening stimuli (Menegas et al., 2018;Menegas et al., 2015). According to 10 projection targets and functional output, the SNc is divided into a medial and lateral part. Medial

90
Next, we analyzed the distribution of Bcl11a-expressing cells using Bcl11a lacZ mice. In this 91 mouse line, the lacZ allele is knocked into the endogenous Bcl11a locus and β-gal expression 92 is restricted to cells that express Bcl11a (Dias et al., 2016). Indeed, the distribution of β-gal 93 positive mDA neurons in neonatal and adult brain was comparable to the one of mDA neurons

106
To investigate the developmental time course of BCL11A expression in the ventral midbrain, 107 we used immunostaining to analyze BCL11A protein expression between embryonic day 6 (E)12.5 and E15.5. BCL11A was first expressed in the ventral midbrain at E12.5. At E12.5 109 and E13.5, BCL11A was mainly localized in the area just below the mDA progenitor domain 110 and in a few differentiated TH-expressing mDA neurons. At E14.5 and E15.5, expression was 111 found in a larger subset of mDA neurons, both in the forming SN and VTA (Supplemental

139
Bcl11a-expressing mDA neurons form a subcircuit in the dopaminergic system 140 Next, we examined whether the Bcl11a-expressing subclass of mDA neurons contributes to 141 specific subcircuits in the mDA system. To investigate projection targets of Bcl11a-expressing 142 mDA neurons we used an intersectional genetic approach. This method combines a reporter 7 allele or viral construct in which expression of a fluorescent reporter protein is driven by a 144 tetracycline response element (TRE) in a Cre-dependent manner (Madisen et al., 2015;Poulin 145 et al., 2018). To achieve specific activation of the reporter allele in mDA neurons we used a 146 Dat tTA (tetracycline trans-activator driven by the Dat promoter) mouse line in conjunction with 147 the Bcl11a CreER mouse line (Chen et al., 2015;Poulin et al., 2018) and the intersectional 148 reporter mouse line Ai82D (Madisen et al., 2015) (Figure 3A,B). Since Bcl11a expression is 149 already restricted to subsets of mDA neurons in the developing brain (Supplemental Figure   150 3B-E) , CreER was activated during embryogenesis by administering Tamoxifen to pregnant 151 females at E15.5. Distribution of the recombined (EGFP+) neurons in the adult brain was  (Lerner et al., 2015;Menegas et al., 2018;Poulin et al., 2018). Within these SN target areas, 163 the densest innervation originating from recombined mDA neurons was observed in the ventral 164 TS. Very sparse innervation was observed in the dorsolateral or rostral striatum.

165
In a next step, we investigated the projections of mDA neurons that express Bcl11a in the adult 166 brain to (1) clarify whether Bcl11a-expression defines the same subset of mDA neurons in the 167 embryonic and adult brain and (2)          neurons has been shown to lead to a decreased preference for social novelty (Bariselli et al., 272 2018), we used a social recognition test to examine the ability of the Bcl11a cko mice to 273 distinguish familiar and unfamiliar mice. We did not find a significant alteration in the behavior 274 of Bcl11a cko mice as compared to control mice (data not shown). Next, we focused on tasks 275 in which dopamine release from SN-mDA neurons is thought to play a prominent role.

280
To examine whether motor skill learning is altered in Bcl11a cko mice, mice had to perform an 281 accelerating rotarod test (Costa et al., 2004). In this task, control mice improved their

389
BCL11A is a zinc finger transcription factor that acts mainly as a repressor but it is also part of 390 an ATP dependent chromatin remodeling complex in neural tissue. Bcl11a is expressed in 391 many types of neurons in almost every region of the CNS (Allen Brain Atlas), but its molecular 392 function has only been studied in a few regions so far (Simon et al., 2020). In the cortex,

393
BCL11A is important for the specification of cortical neurons that project to subcerebral areas 394 (Canovas et al., 2015;Woodworth et al., 2016) and it controls the acquisition of sensory area 395 identity and the establishment of sensory input fields (Greig et al., 2016). Moreover, BCL11A 396 14 regulates the migration of cortical projecting neurons (Wiegreffe et al., 2015). In the dorsal 397 spinal cord, BCL11A is required for neuronal morphogenesis and sensory circuit formation 398 (John et al., 2012). Whether it functions purely as a transcriptional repressor or also as part of 399 the chromatin remodeling complex in these neurons remains to be investigated. In mDA    Bcl11a cko mice show a defect in the learning of skilled motor behavior while spontaneous 436 motor behavior or motor coordination is not affected. Since Bcl11a is specifically inactivated in 437 mDA neurons in our mouse model, this particular behavioral phenotype must be caused by 438 functional changes in mDA neurons (rather than by deficits in the cerebellum or motor cortex 439 (Hikosaka et al., 2002;Li et al., 2017). Due to the highly specific innervation pattern of Bcl11a-440 mDA neurons in the forebrain and the fact that Bcl11a-mDA neurons comprise only a subset 441 of mDA neurons, we were not able to address the nature of these functional changes, but we 442 assume that they ultimately result in altered dopamine release in the target areas of Bcl11a-443 mDA neurons and that these alterations are severe enough to elicit a behavioral phenotype.

444
Learning of motor skills is thought to be mediated by dopamine release in the dorsal striatum.

445
It has been proposed that the plasticity in striatal medium spiny neurons that underlies initial

477
has a more severe dopamine dysfunction than the caudate nucleus (Kish et al., 1988). This

496
This could be due to the particularly high vulnerability of these populations rather than to a 497 specific function of these factors: any additional insult (i.e. loss of ALDH1A1 or BCL11A 498 function) during development or in the adult brain increases their vulnerability even further.

499
Alternatively, BCL11A could be modulating cell survival more directly, since it has been shown

543
For immunofluorescent staining, sections were re-fixed in 4% PFA for 10 min at room 544 temperature (RT) and incubated in 10% NDS in PBS plus 0.2% Triton X-100 (Sigma-Aldrich) 545 (0.2% PBT, used for embryonic and P0 tissue) or in 10% NDS in 0.5% PBT (adult tissue) for 546 1 hr at RT. Sections were incubated with primary antibody overnight at 4°C in 3% NDS in 0.2% 547 PBT (embryonic and P0 tissue) or in 3% NDS in 0.3% PBT (adult tissue). For staining with the 548 guinea pig anti-BCL11A antibody and in some cases for rabbit anti-TH antibody ( Table 1) 549 sections were incubated in the primary antibody for 72 hr at RT (guinea pig anti-BCL11A) or

561
A list of primary and secondary antibodies is provided in Table 1

612
Brightfield images were visualized with a Zeiss Axio Scope.A1 microscope, collected using

614
In situ hybridized sections at adult stages were imaged at an inverted Zeiss AxioObserver 615 equipped with a CSU-W1 Confocal scanner unit (50 μm pinhole disk, Yokogawa, Tokyo, JP).

631
VTA was determined by quantifying TH + β-gal + neurons at four rostrocaudal midbrain levels 632 (Franklin and Paxinos, 2007). The number of TH + β-gal + neurons co-expressing the respective 633 subset marker in these regions was counted unilaterally and the numbers were normalized for 634 the total number of TH + β-gal + mDA neurons in each region (SN or VTA). This analysis was 635 performed for n=5 Bcl11a-lacZ mice and n=5 Bcl11a cko lacZ mice at P0 and for n=3 Bcl11a-  Olympus disk spinning unit (DSU) and a light sensitive EM-CCD camera. Coefficient of error 648 was calculated according to (Gundersen and Jensen, 1987); values <0.10 were accepted.  beam walking assay (balanced beam test) was used (Carter et al., 1999). Animals had to 678 balance from one platform over a 12 mm wide and 1 m long rod made of synthetic material to 679 another platform that held a box with a food reward. Beams were placed 50 cm above the 680 table. The time taken for the animals to cross the bar was measured. If animals did not reach 681 23 the safety platform or took longer than 60 s to cross the beam, a maximum of 60 s was 682 assigned. The test was run for a period of 5 consecutive days.

684
Open-field test

685
In the open-field test, the mice were placed in an open arena (30 x 30 x 30 cm). They were 686 allowed to move freely for 5 minutes. The animals were recorded by Video (EthovisionXT,

687
Noldus, Wageningen, NL) and the running distance, the time spent in the border area, the 688 corners and the centre of the cage as well as crossings from the border to the center area 689 were measured. This test was initially run for a training period of 1 day, then after a 30-day 690 rest period another test run was performed. This test was run twice with a 30-day rest period 691 in between. Since both runs showed comparable results, the results were combined for the 692 final analysis.

694
Statistical analyses of cell numbers were done with GraphPad Prism (8.0) software using 695 unpaired t-test, Welch's t-test, one-way ANOVA followed by test for linear trend or one-way 696 ANOVA followed by Tukey's post hoc test for multiple comparisons.

697
Open field tests were evaluated by one-way analysis followed by Tukey's post hoc test for 698 multiple comparisons. Differences in the balanced beam test and rotarod were assessed by 699 two-way ANOVA taking time and genotype as numerical and categorical variables.