All animal studies were performed in accordance with local animal welfare regulations, as this project has been approved by the animal experimental committee (Dier ethische commissie, Universiteit van Amsterdam, DEC-UvA), and international guidelines.

The Tcf12^flox (Heb^flox) mice strain was a kind gift of Prof Dr. Y. Zhuang from the Department of Immunology of the Duke University Medical Center in Durham, North Carolina (Wojciechowski et al., 2007). Embryos were generated by crossing with En1^cre (Kimmel et al., 2000) and Pitx3^cre (Smidt et al., 2012) mice strains. WT embryos, to investigate normal expression of Tcf12, were generated by crossing C57BL/6 mice. Pregnant (embryonic day 0.5 (E0.5) is defined as the morning of plug formation) and adult mice were sacrificed by cervical dislocation. Embryos and brains were collected in 1× PBS and immediately frozen on dry-ice, or fixed by immersion for 3–12 h in 4% paraformaldehyde (PFA) at 4°C. After PFA incubation, samples were washed in 1× PBS and cryoprotected O/N in 30% sucrose at 4°C. Embryos and brains were frozen on dry-ice and stored at −80°C. Cryosections were sliced at 16 μm, mounted on Superfrost Plus slides (Thermo Fisher Scientific), air-dried, and stored at −80°C until further use.

Genotyping of the Tcf12^flox allele was performed as described (Wojciechowski et al., 2007). PCR was performed with 50 ng of genomic DNA together with FP 5′-CTGGGACAGAAGTTCAGCACTTAGTAC-3′ and RP 5′-CATTCCTATACATCAGCTTCTTGGACG-3′ resulting in a product at 1.1 kb for the WT allele and a product at 1.3 kb for the floxed allele (Supplementary Figure S1A).

Genotyping of the En1^cre allele was performed with 50 ng of genomic DNA together with primer pair En1Cre 5UTR_F3 5′-CTTCGCTGAGGCTTCGCTTT-3′ and En1Cre Cre_R2 5′-AGTTTTTACTGCCAGACCGC-3′ resulting in a product at 240 bp for the cre-allele (Supplementary Figure S1B).

Genotyping of the Pitx3^cre allele was performed as described (Smidt et al., 2012). PCR is performed in two reactions. The first reaction contained 50 ng of genomic DNA together with primer pair iCre-FP2 5′-GCATGATTTCAGGGATGGAC-3′ and iCre-RP2 5′-ATGCTCCTGTCTGTGTGCAG-3′ resulting in a product at 750 bp for the cre-allele. The second reaction contained 50 ng of genomic DNA together with primer pair Pitx3-exon2–3_FP 5′-CAAGGGGCAGGAGCACA-3′ and Pitx3-exon2-3_RP 5′-GTGAGGTTGGTCCACACCG-3′ resulting in a product at 400 bp for the WT allele (Supplementary Figure S1C).

In situ hybridization with digoxigenin (DIG)-labeled probes was performed as described previously (Smidt et al., 2004). Fresh frozen sections were fixed in 4% PFA for 30 min and acetylated with 0.25% acetic anhydride in 0.1 M triethanolamine for 10 min. Probe hybridization was performed at 68°C O/N with a probe concentration of 0.4 ng/μl in a hybridization solution containing 50% deionized formamide, 5× SSC, 5× Denhardt’s solution, 250 μg/ml tRNA Baker’s yeast, and 500 μg/ml sonificated salmon sperm DNA. The following day slides were washed in 0.2 × SSC for 2 h at 68°C followed by blocking with 10% HIFCS in buffer 1 (100 mM TricHCl, pH = 7.4 and 150 mM NaCl) for 1 h at RT. DIG-labeled probes were detected by incubating with alkalin-phosphatase-labeled anti-DIG antibody (Roche, Mannheim), using NBT/BCIP as a substrate. DIG in situ hybridization was performed with the following probes: 965 bp Tcf12 fragment bp 1027–1992 of mouse cDNA, 491 bp Th fragment 252–743 of mouse cDNA, 359 bp Dat fragment bp 762–1127 of rat cDNA, 1796 bp Ahd2 fragment bp 5–1801 of mouse cDNA, 368 bp Cck fragment bp 263–630 of mouse cDNA, 1150 bp fragment Lmx1a bp 218–1366 of mouse cDNA, 2246 bp Nurr1 fragment bp 1–2247 of rat cDNA, and 707 bp Ngn2 fragment bp 98–1002 of mouse cDNA. Slides were washed 2 × 5 min in T₁₀E₅, dehydrated with ethanol, and embedded in Entellan.

After DIG in situ hybridization for Tcf12, E12.5, E14.5, and adult WT sections were immunostained for TH. Slides were incubated in 0.3% H₂O₂ in Tris-buffered saline (TBS) for 30 min at RT. Thereafter, blocking was performed with 4% heat-inactivated fetal calf serum (HIFCS) in TBS. Slides were incubated O/N with primary antibody Rb-TH (Pelfreeze, 1:1000) in TBS. The following day slides were incubated for 1 h with biotinylated goat-anti-rabbit and 1 h with avidin-biotin-peroxidase reagents (ABC Elite kit, Vector Laboratories 1:1000) in TBS. Slides were stained with 3,3′-diamino-benzidine (DAB) for a maximum of 10 min. Slides were dehydrated with ethanol and embedded with Entellan.

Fluorescence immunohistochemistry was performed as described previously (Kolk et al., 2009; Fenstermaker et al., 2010). Cryosections were blocked with 4% HIFCS or 5% normal donkey serum (for sheep primary antibodies) in 1× THZT and incubated O/N with a primary antibody (Rb-TH (Pelfreeze 1:1000), Sh-TH (Millipore AB1542, 1:1000), Rb-PITX3 ((Smidt et al., 2000) 1:1000), Rb-TCF12 (A-20, Santa Cruz sc-357, 1:250), Rb-Ki67 (Abcam ab15580, 1:500), Rb-PhosphoH3 (Abcam ab5176, 1:1000)). The next day sections were incubated with a secondary Alexafluor antibody (anti-rabbit, anti-sheep) diluted 1:1000 in 1× TBS for 2 h at RT. After washing with 1× PBS, slides were incubated with DAPI (1:3000) 5 min at RT and after extensive washing with 1× PBS embedded in Fluorsave (Calbiogen). Analysis was performed on a fluorescent microscope.

DAB immunohistochemistry on adult sections was performed with Rb-AHD2 (Abcam ab23375 1:1000) as described above, with some alterations. The AHD2 antibody required antigen retrieval as follows. Cryosections were incubated with 0.1 M citrate buffer pH6 for 3 min at 800 W and 9 min a 400 W, cooled down to RT in a water bath. Thereafter slides were blocked in 4% HIFCS in 1× THZT and incubated O/N with a primary antibody. The following day slides were incubated for 1 h with biotinylated goat-anti-rabbit and 1 h with avidin-biotin-peroxidase reagents (ABC Elite kit, Vector Laboratories 1:1000) in TBS. The slides were stained with DAB for a maximum of 10 min. Dehydration was performed with ethanol and embedding with Entellan.

Quantification of TH-, PITX3- and P-H3-expressing cells was performed in ImageJ as follows. TH- and PITX3-expressing cells were counted in 5–10 (matching) sagittal sections of E12.5 and E14.5 embryos (n = 3 Tcf12^+/+;En1^cre/+/Tcf12^+/+; Pitx3^cre/+ n = 3 Tcf12^lox/lox;En1^cre/+/Tcf12^lox/lox;Pitx3^cre/+). P-H3 expressing cells were counted in 5 (matching) coronal sections of E12.5 embryos (n = 3 Tcf12^+/+;En1^cre/+ n = 3 Tcf12^lox/lox;En1^cre/+). Cells were counted as positive, if staining co-localized with a nuclear DAPI staining.

TH- and AHD2-expressing cells in the adult brain were quantified in 6–11 (matching) coronal sections of P30 adult brains (n = 3 Tcf12^+/+;En1^cre/+/Tcf12^+/+;Pitx3^cre/+ n = 3 Tcf12^lox/lox;En1^cre/+/Tcf12^lox/lox;Pitx3^cre/+). The SNc and VTA were determined based on anatomical landmarks for quantification of the TH-expressing cells. The SNc was clearly determined in sections rostral from the fasciculus retroflexus, whereas the mdDA neuronal population was divided in the VTA and SNc in sections containing and caudal from the fasciculus retroflexus. The distinction between the SNc and VTA was made based on the tracts of the medial lemniscus, positioned in between the SNc and VTA. TH-expressing cells were counted as positive when staining co-localized with nuclear DAPI staining. Quantification of AHD2⁺ cells performed on DAB-immunohistochemistry cells were counted as positive when they showed specific DAB staining in the cytoplasm.

Quantifications are expressed as percentages of the WT (with WT set at 100%) ± standard deviation. Parametric statistical analysis was performed through a student’s t-test. p < 0.05 was considered significant, and indicated using an asterisk (*).

As introduced above, Tcf12 is a member of the E-box protein family, a sub-family of the bHLH protein family. The bHLH protein family-members Ngn2 and Mash1 have been implicated in the development of mdDA neurons (Andersson et al., 2006a; Kele et al., 2006). To determine whether Tcf12 is expressed in the midbrain during mdDA development, we characterized its expression pattern by means of in situ hybridization (Figure 1A). Tcf12 is expressed in the VZ of the midbrain and shows a strong expression in the FP at E11.5 and E12.5. At E12.5, Tcf12 expression overlaps with the TH-expressing population in the most medial sections of the midbrain. At E14.5 the expression of Tcf12 becomes further restricted and, similarly to E12.5, overlaps with the most medial part of the TH-expressing population. In adult mdDA neurons Tcf12 transcript could not longer be detected.

In order to confirm the expression pattern of Tcf12, we performed immunohistochemistry for TCF12 protein in combination with TH at E12.5 and E14.5. TCF12 protein is present in a similar pattern as Tcf12 transcript and in concordance co-localizes with part of the medial TH-expressing population (Figure 1B). At E12.5, TCF12 is present in the VZ of the midbrain, and is highly expressed in the FP. Co-localization with TH is detected in both caudally and rostrally (Figures 1B1,3 respectively, white arrowheads) located TH⁺ neurons in the medial part of the midbrain. At E14.5 the expression of TCF12 becomes even more restricted to the medial part of the midbrain and shows co-localization with caudally localized TH⁺ neurons (Figures 1B2,4, white arrowheads).

Together, these data show that Tcf12 transcript and TCF12 protein are present within the midbrain region and co-localize with part of the TH-expressing neuronal population at E12.5 and E14.5. TH⁺/TCF12⁺ neurons are mostly located in the caudal-medial part of the midbrain area. Since TH-expressing neurons are thought to arise from the FP and BP from E10 onwards and then migrate to more lateral and rostral parts of the midbrain (Ono et al., 2007; Bye et al., 2012; Mesman et al., 2014), the expression pattern of TCF12 suggests that co-localization with TH only occurs in relative young TH⁺ neurons and that Tcf12 may serve a function in the initial phases of differentiation of neural progenitors towards the mdDA neuronal phenotype.

As shown above, expression of Tcf12 is mainly present in the VZ and shows co-localization with relative young TH⁺ neurons, indicating that it coincides with the area in the midbrain that generates mdDA neurons. To confirm that Tcf12 could serve an early function in mdDA neuronal development, we examined whether the expression of Tcf12 resembles the expression of known early specification factors in the midbrain (Figure 2A).

Expression of Tcf12 overlaps almost completely with the expression pattern of Lmx1a and partly with Ngn2, both important in the early specification of mdDA neurons (Andersson et al., 2006a; Kele et al., 2006; Yan et al., 2011), at E12.5 (adjacent sections). Both factors show a slightly larger area of expression than Tcf12, mainly in the ventral region of the expression area of Tcf12. Nurr1, an important intrinsic differentiation factor of mdDA neurons (Saucedo-Cardenas et al., 1998; Andersson et al., 2007; Kim et al., 2007), on the other hand, overlaps only minimally at the medial parts of Tcf12 expression at E12.5 (adjacent sections). This expression pattern supports the idea that Tcf12 is mainly involved in relative early events of the development of mdDA neurons.

The expression pattern of Tcf12 combined with the expression patterns of early specification factors strongly suggests a role for Tcf12 in relative early development of mdDA neurons. In order to examine the effect of early loss of Tcf12 on the expression of these factors we crossed the Tcf12^flox with the En1^cre driver (Tcf12^lox/lox;En1^cre/+). En1 is broadly expressed in the midbrain and R1 from ~E9 onwards (Wurst et al., 1994; Kouwenhoven et al., 2016), resulting in a broad and relative early loss of Tcf12. En1^cre driven loss of Tcf12 does not display a difference in the expression of Ngn2 in the midbrain area at E12.5 (Figure 2B, upper panel), whereas in adjacent sections we detected a spatial difference in the expression pattern of both Lmx1a (Figure 2B, middle panel) and Nurr1 (Figure 2B, lower panel). This effect was seen in at least 2 different litters. The Lmx1a expression area is smaller in the midbrain of Tcf12^lox/lox;En1^cre/+ embryos compared to Tcf12^+/+;En1^cre/+ embryos and lacks lateral expression (Figure 2B, middle panel). The expression of Nurr1 shows a decrease in the area of expression, and similar to the expression of Lmx1a, the main effect on the expression of Nurr1 can be detected in the lateral parts (Figure 2B, lower panel). Together, these data suggest that Tcf12 influences the expression of other essential early specification factors in mdDA neuronal development.

Above we have shown that En1^cre driven deletion of Tcf12 results in a spatial difference in expression of Lmx1a and Nurr1, which are both essential determinants of mdDA neuronal differentiation and subset specification (Saucedo-Cardenas et al., 1998; Andersson et al., 2006b, 2007; Kim et al., 2007; Hoekstra et al., 2013). Since Nurr1 is expressed in mdDA neurons and deletion of Nurr1 results in a complete loss of TH-expression in mdDA neurons (Saucedo-Cardenas et al., 1998; Smits et al., 2003), we examined the effect of the loss of Tcf12 on the appearance of mdDA neurons in the Tcf12^lox/lox;En1^cre/+ mutant at E12.5 and E14.5, by examining the spatio-temporal protein levels of the mdDA neuronal markers TH and PITX3. At E12.5 the presence of TH (Figures 3A1,2) and PITX3 (Figures 3A3,4) is decreased with ~66% (n = 3; p < 0.001, one-tailed t-test; Figure 3C E12.5) and ~82% (n = 3; p < 0.001, one-tailed t-test; Figure 3D E12.5; Absolute numbers are represented in Supplementary Figures S2A–D), respectively. At E14.5 the loss of TH-expressing neurons has recuperated (Figure 3B) to a ~33% decrease (n = 3; p < 0.001, one-tailed t-test; Figure 3C E14.5), whereas the loss of PITX3⁺ neurons is recovered at E14.5 (n = 3; Figures 3B,3,4,D, E14.5). The decrease in TH⁺ neurons at E14.5 can mainly be detected in the rostral parts of the TH-expressing population (Figures 3B,1,2). Double-labeling of TH and PITX3 at E14.5 shows that the amount of TH⁺/PITX3⁺ neurons in the total PITX3-expressing population is ~68% in Tcf12^+/+;En1^cre/+ embryos and has decreased to ~50% in the Tcf12^lox/lox;En1^cre/+ embryos (n = 3; p < 0.01 one-tailed t-test; Figure 3E), showing that the population TH⁺/PITX3⁺ neurons has decreased in the mutant. This effect is most prominent in the rostral population of PITX3-expressing neurons (Supplementary Figure S3).

Together, these data show that deletion of Tcf12 in the En1 expression domain results in a delay of expression of the mdDA markers TH and PITX3, which partly (TH) or even completely (PITX3) recovers during development.

Since Tcf12 is known to be important in regulation of proliferation and differentiation of T-cells in the immune system and possibly cortical neurons (Uittenbogaard and Chiaramello, 2002; Wojciechowski et al., 2007) and we detected a delay in mdDA neuronal specification, we examined the effect on En1^cre driven loss of Tcf12 on the proliferation of neural progenitors in the VZ. MdDA neurons are born from E10 onwards, with neuronal birth peaking at E11-E12 (Bayer et al., 1995; Bye et al., 2012). If the apparent delay in mdDA neuronal differentiation is the result in a delay of proliferation, we would expect that the level of cells that are positive for PhosphoH3 (P-H3) and/or Ki67, both markers for proliferation, would be increased in the mutant animals at E12.5. However, we were not able to detect a clear increase in either P-H3 (Figures 4A1,2,B, Supplementary Figure S2E) or Ki67 (Figures 4A3,4) in the midbrain area between the Tcf12^lox/lox;En1^cre/+ mutant and Tcf12^+/+;En1^cre/+ embryos at E12.5, suggesting that the effects found on TH and PITX3 protein are not a consequence of large changes in proliferation.

As described above, we detected a delay in the development of mdDA neurons unrelated to proliferation defects. This suggests that the differentiation of mdDA neurons is affected and the underlying genetic programming may be altered as well. Since rostral TH expression is still affected at E14.5, whereas the presence of PITX3 is restored, we examined the expression of Ahd2, a marker for the rostral mdDA neuronal population and involved in RA dependent Th gene activation (Jacobs et al., 2007, 2011), and Cck, a marker for the caudal mdDA neuronal population (Veenvliet et al., 2013), in order to validate the programming of emerging mdDA neurons in these mutants. Expression of Ahd2 is strongly decreased (Figure 4C, upper panel), whereas the expression of Cck is relatively unaffected at E14.5 in these mutants (Figure 4C, lower panel), suggesting that rostral mdDA neurons do not only lose expression of TH, but also do not express Ahd2 at this stage. Taken together, the presented data suggest that Tcf12 causes a delay in programming and affects the rostral subgroup by an absence of Ahd2 expression at E14.5. To further substantiate the initial programming defects observed at E14.5 we examined the expression of Ahd2 and Cck at E16.5 and of Th, Dat, Ahd2, and Cck at P30 (Figure 5). At E16.5 expression of Ahd2 (Figure 5A, left panel) is partially recovered compared to its expression at E14.5. However, the expression pattern is still affected compared to controls. Similar as described for E14.5, Cck is normally expressed in the E16.5 embryonic midbrain (Figure 5A, right panel).

In the adult midbrain expression of the mdDA neuronal markers Th (Figure 5B, upper left panel) and Dat (Figure 5B, lower left panel) show a modest loss in expression in the lateral and caudal parts of the SNc (Figure 5B, black arrowheads). Interestingly, besides a loss of Dat and Th expression, the overall level of these markers appears to be affected in caudal parts of the midbrain, whereas in more medial parts of the SNc, an increase in Dat transcript level can be detected (black arrows). Expression of the subset markers Ahd2 (Figure 5B, upper right panel) and Cck (Figure 5B, lower right panel 1,2) appears to be increased in adjacent sections (Figure 5B, black arrows). The increase in expression of these subset markers is mainly detected in medial parts of the midbrain. These changes in transcript expression were seen in several adult animals.

To substantiate the expression data as described above, we aimed to quantify the amount of TH- and AHD2-expressing neurons in the adult mdDA neuronal population. In accordance to Th transcript, TH-expressing cells are specifically lost in the SNc region (Figures 6A1–4). The total amount of TH-expressing cells is ~14% lower in Tcf12^lox/lox;En1^cre/+ (n = 3; p < 0.05, one-tailed) than in Tcf12^+/+;En1^cre/+ littermates (Figure 6B, Supplementary Figure S2F). The reduction in TH-expressing cells is only observed in the SNc where the amount of TH⁺ neurons is reduced with ~31% (n = 3; p < 0.05, one tailed), whereas in the VTA the amount of TH⁺ neurons is not significantly changed in Tcf12^lox/lox;En1^cre/+ animals (n = 3; Figure 6B). Similar to the increased Ahd2 transcript expression area, an increase in AHD2-expressing cells (Figures 6C1–4) of ~10% (n = 3; p < 0.05, one-tailed) in Tcf12^lox/lox; En1^cre/+ animals (Figure 6D, Supplementary Figure S2G) is detected. Together, these data show that the population of TH-expressing neurons is specifically affected in the SNc in the adult stage. The observed increase in AHD2-containing neurons within the mdDA neuronal population suggests that the subset programming of the remaining mdDA neurons in the mutant is affected as well.

Above we have shown that early deletion of Tcf12 under the En1^cre driver results in a loss of TH expression in the SNc and overall programming defects. The latter suggests that Tcf12 may influence terminal differentiation of mdDA neurons. In order to validate this notion we created a Pitx3^cre driven conditional ablation of Tcf12.

Contrary to a relative early deletion of Tcf12, Pitx3^cre driven deletion of Tcf12 does not affect TH expression in mdDA neurons at E14.5 (Figure 7A, Supplementary Figure S2H, n = 3). Interestingly, expression of the subset marker Ahd2 is similarly lost in the Tcf12^lox/lox;Pitx3^cre/+ mutant at E14.5 (Figure 7B, left panel), whereas the spatial expression of Cck appears to be diminished in the rostral-medial expression area (Figure 7B, right panel). These data corroborate the initial finding that Tcf12 does influence the programming of mdDA neurons in the terminal differentiation state.

In order to follow the development of the system and confirm these programming defects in adult mutants, we quantified the amount of TH- (Figure 8A) and AHD2-expressing cells (Figure 8C). Quantification of the total amount of TH-expressing neurons and in the SNc and VTA separately, shows, in concordance to earlier time-points, that the amount of TH neurons is unaffected by late ablation of Tcf12 (Figures 8A1–4,B, Supplementary Figure S2I, n = 3). Noteworthy, although Ahd2 expression is completely lost in Tcf12^lox/lox;Pitx3^cre/+ embryos at E14.5, this effect is absent at P30 (Figures 8C1–4,D, Supplementary Figure S2J, n = 3). Not only could this be detected at the protein level, also transcript expression of Th, Ahd2 and Cck in the adult midbrain was unaffected in Tcf12^lox/lox;Pitx3^cre/+ animals (Supplementary Figure S4).

Taken together these data show that Pitx3^cre driven deletion of Tcf12 confirms the temporal delay in correct coding of mdDA neurons, which, in late ablation of Tcf12, is corrected in the adult system.