Plag1 and Plagl2 have overlapping and distinct functions in telencephalic development

ABSTRACT The Plag gene family has three members; Plagl1/Zac1, which is a tumor suppressor gene, and Plag1 and Plagl2, which are proto-oncogenes. All three genes are known to be expressed in embryonic neural progenitors, and Zac1 regulates proliferation, neuronal differentiation and migration in the developing neocortex. Here we examined the functions of Plag1 and Plagl2 in neocortical development. We first attempted, and were unable to generate, E12.5 Plag1;Plagl2 double mutants, indicating that at least one Plag1 or Plagl2 gene copy is required for embryonic survival. We therefore focused on single mutants, revealing a telencephalic patterning defect in E12.5 Plagl2 mutants and a proliferation/differentiation defect in Plag1 mutant neocortices. Specifically, the ventral pallium, a dorsal telencephalic territory, expands into the ventral telencephalon in Plagl2 mutants. In contrast, Plag1 mutants develop normal regional territories, but neocortical progenitors proliferate less and instead produce more neurons. Finally, in gain-of-function studies, both Plag1 and Plagl2 reduce neurogenesis and increase BrdU-uptake, indicative of enhanced proliferation, but while Plagl2 effects on proliferation are more immediate, Plag1 effects are delayed. Taken together, we found that the Plag proto-oncogenes genes are essential regulators of neocortical development and although Plag1 and Plagl2 functions are similar, they do not entirely overlap. This article has an associated First Person interview with the first author of the paper.

All three members of the Plag gene family encode zinc finger transcription factors that share homology chiefly in their amino terminal zinc (Zn) finger domains, whereas the carboxyl terminal regions of the three proteins are quite diverse . Several transcriptional targets of the Plag family transcription factors have been identified. For example, Plag1 and Plagl2 both regulate the expression of Insulin-like Growth Factor 2 (Igf2), which accounts at least in part for their abilities to stimulate cell proliferation Declercq et al., 2008;Varrault et al., 1998;Voz et al., 2000). In addition, Plag1/Plagl2 promote tumorigenesis by initiating the transcription of several Wnt pathway genes. For instance, Plagl2 has been shown to regulate expression of Wnt6, Fzd2 and Fzd9 to maintain cells in a proliferative state (Zheng et al., 2010). Likewise, Plag1 misexpression in pleomorphic adenomas results in an upregulation of canonical Wnt signaling Zhao et al., 2006). Finally, Plag1 was also found to regulate several cell division and cell cycle-related genes, such as Cyclin D3 and Cyclin D1, as well as apoptosis-related genes, such as Caspase-8 (Voz et al., 2004).
Despite extensive knowledge of Plag gene function in cancer, their roles during normal development have only recently been examined. Zac1, Plag1 and Plagl2 all function to regulate embryonic growth Van Dyck et al., 2007a;Varrault et al., 2006). Zac1 also controls development of keratinocytes (Basyuk et al., 2005), heart (Czubryt et al., 2010;Yuasa et al., 2010) and pancreatic islets (Anderson et al., 2009), while Plagl2 functions to control the development of enterocytes (Van Dyck et al., 2007a). All three Plag genes are expressed in several lineages in the developing embryo as well as in some adult tissues, each with unique expression domains that overlap in certain lineages/tissues (Alam et al., 2005;Hensen et al., 2004;Rodríguez-Henche et al., 2002;Van Dyck et al., 2007a). For example, Zac1 is expressed in a regionalized fashion in neural progenitor cells in the developing central (CNS) and peripheral (PNS) nervous systems, whereas Plag1 and Plagl2 are more uniformly expressed in CNS and PNS neural progenitors (Abdollahi, 2007;Alam et al., 2005;Astrom et al., 1999;Poulin and Labelle, 2005). Interestingly, all three Plag genes are co-expressed at higher levels in neural progenitors than in post-mitotic neurons (Alam et al., 2005;Rodríguez-Henche et al., 2002).
Plag1 null mice (Plag1 KI/KI ), although viable, are growth retarded and have reduced fertility . However, despite their growth defects and the known ability of Plag1 to regulate expression of the Igf2 growth factor (Voz et al., 2000), Igf2 expression levels were found to be unperturbed in Plag1 null mice . Thus, the underlying molecular mechanisms that lead to growth perturbation in Plag1 null embryos remain unknown. Likewise, Plagl2 mutant neonates also weigh less relative to their littermates at birth (Van Dyck et al., 2007a). However, unlike Plag1 mutants, Plagl2 mutant pups display postnatal lethality, dying shortly after birth due to starvation and nutrient malabsorption (Van Dyck et al., 2007a). In the neonatal Plagl2 mutant liver, the starvation response factor asparagine synthetase is expressed at high levels (Van Dyck et al., 2007a), whereas Igf1 levels are low, indicative of a loss of nutrients.
In the CNS, multiple developmental roles for Zac1 have been deciphered, including in the retina, cerebellum and neocortex Chung et al., 2011;Ma et al., 2007a,b;Rraklli et al., 2016). However, to date, neither Plag1 nor Plagl2 have any known functions in the developing CNS. Here, given their overlapping expression with Zac1, we asked whether Plag1 and Plagl2, also function during neocortical development, revealing novel and specific roles for these genes in both telencephalic patterning and in regulating neocortical progenitor cell proliferation and neurogenesis.

RESULTS
Plag1 and Plagl2 do not cross-regulate each other at the level of transcription Plag1 and Plagl2 have similar amino acid sequences, sharing 79% and 35% identity in their N-and C-termini, respectively (Juma et al., 2016). They also share several transcriptional targets, including the growth factor Igf2 (Abdollahi, 2007). In addition, Plag1 and Plagl2 have both been characterized as growth regulators and proto-oncogenes (Abdollahi, 2007;Juma et al., 2016;Landrette et al., 2005). Here we set out to determine whether they also have overlapping and possibly redundant roles in the developing telencephalon.
To better understand how Plag1 and Plagl2 function in the embryonic telencephalon, we first examined their expression profiles at embryonic day (E) 12.5, when the first neurons have begun to differentiate in both dorsal and ventral domains (Adnani et al., 2018). As previously reported (Alam et al., 2005), Plag1 (Fig. 1A) and Plagl2 (Fig. 1C) were expressed in E12.5 telencephalic progenitors throughout the dorsal and ventral ventricular zones (VZ) in a highly similar fashion, albeit with apparently higher Plagl2 transcript levels. We also monitored the expression of these genes by taking advantage of the knockin of lacZ into the Plag1 [Plag1 lacZKI/+ ; Fig. 1E; Hensen et al. (2004)] and Plagl2 [Plagl2 lacZKI/+ ; Fig. 1F; Van Dyck et al. (2007a)] loci. X-gal staining of coronal sections of E12.5 Plag1 lacZKI/+ (hereafter Plag1 KI/+ ; Fig. 1G) and Plagl2 lacZKI/+ (hereafter Plagl2 KI/+ ; Fig. 1H) heterozygous brains revealed that lacZ had a similar distribution throughout the telencephalic VZ in both genotypes, again with apparently higher levels of lacZ expression in Plagl2 KI/+ cortices. Plag1 and Plagl2 are thus expressed similarly in the early embryonic telencephalic VZ, an expression profile that we previously demonstrated persists into the late embryonic period (Alam et al., 2005).
In several instances, highly related genes in the same family are not only expressed in the same CNS domains, but also display crossregulatory interactions. For example, the highly similar proneural genes Neurog1 and Neurog2 are largely co-expressed in the early embryonic dorsal telencephalon (Britz et al., 2006;Han et al., 2018), and Neurog2 is required to initiate Neurog1 expression in the dorsomedial cortex (Fode et al., 2000). To determine whether there are similar cross-regulatory interactions between Plag1 and Plagl2 in the developing telencephalon, we asked whether mutation of one gene can alter expression of the other gene. By RNA in situ hybridization, we observed a similar distribution of Plag1 transcripts in the telencephalic VZ of E12.5 Plagl2 KI/KI homozygous mutants as seen in wild-type brains (Fig. 1A,B). The converse was also true, as Plagl2 transcripts were maintained in the telencephalic VZ of E12.5 Plag1 KI/KI homozygous mutants in a similar pattern as in wild-type brains (Fig. 1C,D).
To validate that there were no differences in Plag1 and Plagl2 transcript levels in homozygous mutants for the other Plag gene, we micro-dissected out telencephalic tissue at E12.5 and performed RT-qPCR (Fig. 1I). As expected, Plag1 transcripts were not detectable in Plag1 KI/KI cortices, whereas Plag1 transcript levels were at the same relative level in wild-type controls and in E12.5 Plagl2 KI/KI homozygous mutant cortices (Fig. 1J). Similarly, Plagl2 transcripts were not detectable in cortical tissue from Plagl2 KI/KI homozygous mutants, whereas Plagl2 transcripts were expressed at wild-type levels in E12.5 Plag1 KI/KI cortices (Fig. 1J′). To complete this data set, we also analyzed expression levels of the third member of this gene family, Plagl1 (also known as Zac1), which differs in that it is a tumor suppressor gene (Abdollahi, 2007). Zac1 is expressed at high levels in neocortical progenitors and plays a role in regulating neuronal morphology and migration . While no differences in Zac1 transcript levels were observed in Plag1 KI/KI cortices, there was a small but significant reduction in Zac1 expression in Plagl2 KI/KI dorsal telencephalic tissue (Fig. 1J″, P<0.01, n=3). Finally, a comparison of relative transcript levels (normalized to the same housekeeping genes) revealed that Zac1 is expressed at the highest levels in cortical cells, followed by Plagl2 and Plag1.
Thus, all three Plag genes are expressed in cortical progenitors, and while there is no evidence of cross-regulatory transcriptional interactions between Plag1 and Plagl2, Zac1 levels are reduced in Plagl2 mutants, at least at E12.5, which is the stage we focused on for the remainder of this study.
Plag1 and Plagl2 act redundantly to control embryonic survival Previous reports have suggested that Plag1 KI/KI null mice are viable after birth, but are growth retarded and have reduced fertility . Likewise, Plagl2 KI/KI neonates weigh less relative to their littermates at birth (Van Dyck et al., 2007a). However, unlike Plag1 KI/KI mutants, Plagl2 KI/KI pups display  postnatal lethality, dying shortly after birth due to starvation and nutrient malabsorption (Van Dyck et al., 2007a). To determine whether Plag1 and Plagl2 function redundantly or have distinct functions in the embryonic telencephalon, we set out to generate double mutants by setting up double heterozygous intercrosses between Plag1 KI/+ ;Plagl2 KI/+ male and female mice. We collected seventeen litters at E12.5 for a total of 120 live embryos and compared the acquired genotypes to the expected genotypes using a Mendelian Punette square diagram for a dihybrid cross (Fig. 1L). If there was no embryonic lethality, we expected Mendelian ratios for each possible genotype after double heterozygous intercrosses. Of the 120 embryos genotyped, significantly under-represented genotypes included Plag1 +/+ ;Plagl2 KI/KI , Plag1 KI/+ ;Plagl2 KI/KI and Plag1 KI/KI ;Plagl2 KI/KI (Fig. 1M).
As we did not collect any double mutant embryos, our data suggests that Plag1 and Plagl2 function redundantly to control embryonic survival. Moreover, the Plagl2 KI/KI genotype has an early embryonic lethal phenotype. This finding was somewhat surprising given that Plagl2 KI/KI mutant embryos were previously reported to survive postnatally (Van Dyck et al., 2007a). Differences between the two studies are likely related to our use of a different genetic background (i.e. CD1) compared to 129/SvJ background used by Van Dyck et al. (2007a).
Plagl2 is required to set the positioning of ventral gene expression at the pallial-subpallial border Neuronal fate specification is directly linked to dorsoventral regional identity in the telencephalon, with progenitors in the dorsal telencephalon giving rise to glutamatergic excitatory projection neurons, while ventral progenitors give rise to GABAergic inhibitory interneurons (Schuurmans and Guillemot, 2002). Given that Plag1 and Plagl2 were expressed in both dorsal and ventral telencephalic progenitors, we first asked whether the Plag genes acted upstream of regional patterning genes. Genes involved in the initial patterning of telencephalic domains are expressed in a regionalized manner and are enriched in, or restricted to, precise dorsal or ventral telencephalic progenitor domains, displaying sharp dorsoventral boundaries (Hoch et al., 2009). Mutation of several patterning genes can disrupt the positioning of borders between dorsal and ventral domains, ultimately affecting the generation of the brain territories derived from these regionalized progenitors. For instance, mutation of the homeobox gene Gsx2 results in an expansion of dorsal telencephalic ( pallial) territories and a corresponding reduction in ventral telencephalic (subpallial) territories, whereas Pax6 mutants have the opposite phenotype (Yun et al., 2001).
To assess the roles of the Plag genes in dorsoventral patterning, we focused on Plag1 and Plagl2 single mutants as we were not able to generate E12.5 Plag1;Plagl2 double mutants. In these embryos, we compared the position of the dorsoventral border between high and low gene expression in the telencephalon (red arrowheads; Fig. 2C-K) to a morphological landmark, the corticostriatal angle (black arrowheads; Fig. 2C-K). To provide quantitative measurements of regional differences, we measured the length ( Fig. 2A) and angle ( Fig. 2B) between the corticostriatal border (morphological landmark) and the pallial-subpallial boundary (PSPB, gene expression landmark), using a fixed lever length for angle measurements.
We first examined the expression of Ascl1, a proneural gene encoding a basic-helix-loop-helix (bHLH) transcription factor that is required for ventral telencephalic development (Casarosa et al., 1999). At E12.5, Ascl1 was expressed at high levels in the ventral telencephalic VZ, including in both the lateral (LGE) and medial (MGE) ganglionic eminences (Fig. 2C), as previously reported (Casarosa et al., 1999). Ascl1 transcripts were also enriched in the cortical hem, while lower levels of Ascl1 transcripts were detected in the dorsal telencephalic VZ (Fig. 2C). The dorsal most limit of the high Ascl1-expression domain marked the PSPB; immediately ventral to the PSPB was an Ascl1-high subpallial territory known as the dorsal LGE (dLGE) and immediately dorsal to the PSPB was an Ascl1-low pallial territory known as the ventral pallium (Yun et al., 2001) (Fig. 2C). A similar pattern of expression was observed in E12.5 Plag1 KI/KI mutant brains (Fig. 2D), while the high Ascl1 expression domain appeared to shift ventrally in E12.5 Plagl2 KI/KI cortices (Fig. 2E). Indeed, measurement of the length (n=3; P<0.01; Fig. 2L) and angle (n=3; P<0.01; Fig. 2M) between the corticostriatal angle and the limit of the high Ascl1 expression domain confirmed that these values were greater in the E12.5 Plagl2 telencephalon compared to wild-type brains.
To provide further validation for this finding, we also examined the expression of Dlx1, a homeodomain transcription factor that acts with the related gene Dlx2 to establish a ventral telencephalic identity, with the absence of these genes, resulting in the loss of most if not all GABAergic interneurons (Anderson et al., 1997). In E12.5 wild-type brains, Dlx1 was expressed at high levels throughout the VZ of the LGE and MGE, with its dorsal limit in the dLGE, and no expression in the ventral pallium ( Fig. 2F). A similar pattern of expression was observed in the E12.5 Plag1 KI/KI telencephalon ( Fig. 2G), whereas in E12.5 Plagl2 KI/KI brains ( Fig. 2H), the dorsal limit of the Dlx1 expression domain was positioned more ventrally.
Taken together, these data suggest that Plagl2 is required to maintain the position of the PSPB in the E12.5 telencephalon, a contention that we investigated further with additional marker analysis.
Plagl2 is required to set the positioning of dorsal gene expression at the pallial-subpallial border We next examined the expression of dorsally-restricted genes in E12.5 Plag1 and Plagl2 mutants. The proneural gene Neurog2, which also encodes a bHLH transcription factor, is required to specify a dorsal telencephalic identity (Fode et al., 2000). In the E12.5 wild-type telencephalon, Neurog2 was exclusively expressed in the dorsal telencephalic VZ, with the ventral border of expression extending into the ventral pallium and ending at the PSPB (Fig. 2I). A similar pattern of Neurog2 expression was observed in the E12.5 Plag1 KI/KI (Fig. 2J) telencephalon, while the limit of Neurog2 expression extended ventrally in Plagl2 KI/KI brains (Fig. 2K). We validated these observations by performing measurements of the length (n=3; P<0.01; Fig. 2N) and angle (n=3; P<0.01; Fig. 2O) between the corticostriatal angle and the limit of high Neurog2 expression.
To further examine the position of the PSPB, we performed costaining of a dorsal pallial marker, the homeodomain transcription factor Pax6, and a ventral marker, the homeodomain transcription factor Gsx2 (Fig. 3). While Pax6 expression extends into the Gsx2 + dLGE territory at E10.5, resulting in the co-expression of these two transcription factors in the dLGE (Yun et al., 2001), we found that at E12.5, Pax6 and Gsx2 expression domains directly abutted one another at the PSPB; Pax6 was expressed at high levels throughout the pallium, including in the ventral pallium, whereas Gsx2 was expressed exclusively in the LGE and MGE, including in the dLGE (Fig. 3A,D-D‴). A similar pattern of co-expression was observed in E12.5 Plag1 KI/KI telencephalons; high Pax6 expression in the pallium and Gsx2 expression restricted to the LGE/MGE, with little to no overlap at the PSPB (Fig. 3B,E-E‴). In contrast, in the Plagl2 KI/KI telencephalon Pax6 expression extended further ventrally, while the Gsx2 expression limit was restricted to a more ventral position, but there was still limited overlap between these markers (Fig. 3C,F-F‴). To confirm these apparent differences in expression, we quantitated the distance between the corticostriatal angle and the border between the Gsx2 and Pax6 expression domains ( presumptive PSPB; Fig. 3G), revealing that this distance was indeed larger in the Plagl2 KI/KI telencephalon (n=3; P<0.05; Fig. 3H).

Expansion of the ventral pallium in Plagl2 mutants
The shift in the positioning of the PSPB could mean that there was an expansion in the size of the ventral pallium, and possibly a corresponding reduction in the size of the dLGE, as observed in Gsx2 mutants (Yun et al., 2001). To test this possibility more fully, we examined the expression of markers that specifically label these two regional territories, using Dbx1 to label the ventral pallium (Bielle et al., 2005), and Etv1/ER81 (Yun et al., 2001) and Sp8 (Waclaw et al., 2006) to mark the dLGE. In E12.5 wild-type cortices, Dbx1 labeled a small stripe of cells ventral to the corticostriatal angle, which is the ventral pallium (Fig. 4A,A′). The Dbx1 + ventral pallial territory was of a similar size and position in E12.5 Plag1 KI/KI brains (Fig. 4B,B′). In contrast, in E12.5 Plagl2 KI/ KI brains, Dbx1 expression initiated in the same position just ventral to the corticostriatal angle, but it extended further ventrally, consistent with the idea that the ventral pallium is expanded in size in Plagl2 mutants (Fig. 4C,C′). Finally, to determine whether there was a compensatory decrease in the size of the dLGE, we examined the expression of Sp8, which marked a thin, similarly sized ventral stripe of cells in both E12.5 wild-type (Fig. 4D,D′) and Plagl2 KI/KI (Fig. 4E,E′) brains. In addition, no noticeable differences were observed in the Etv1 + dLGE domain size in E12.5 wild-type (Fig. 4F,F′) and Plagl2 KI/KI (Fig. 4G,G′) brains.
We first focused on the Tbr1 + mantle zone in the presumptive piriform cortex, which extended in a wedge shape from the pial surface towards the limit of the SVZ at the corticostriatal angle in E12.5 wild-type embryos (Fig. 5D,D′,H,H′). Similar patterns of Tbr1 expression were observed in the mantle of the piriform cortex in E12.5 Plag1 KI/KI (Fig. 5E,E′) and Plagl2 KI/KI (Fig. 5F,F′) brains, with no reduction in the total number of Tbr1 + cells (n=3; P>0.05 for all comparisons; Fig. 5I).
Taken together, these data suggest that Plagl2 is required to maintain the position of the border between the ventral pallium and dLGE at the primary VZ progenitor and secondary SVZ progenitor stage. While there was no obvious reduction in the size of the Tbr1 + neuronal pool in the mantle of the presumptive piriform cortex, the complex migration patterns of ventral pallium-derived neurons, which include Cajal-Retzius neurons that migrate tangentially to cover the pallial surface (Bielle et al., 2005), makes it difficult to specifically assess neuronal output from the ventral pallium. Plag1 is required to regulate the differentiation and proliferation of early embryonic neocortical progenitors To further assess the functions of the Plag genes in the developing telencephalon, we next focused on dorsal pallial territories, which give rise to the neocortex. To first ask whether Plag1 and Plagl2 were required to regulate the proliferation of neocortical progenitors, we administered BrdU 30 min prior to sacrifice to label progenitor cells in S-phase of the cell cycle. In E12.5 wild-type (Fig. 6A), Plag1 KI/KI (Fig. 6B) and Plagl2 KI/KI (Fig. 6C) cortices, BrdU was detected in an abventricular band where S-phase progenitors accumulate due to interkinetic nuclear migration. Quantitation revealed that there was a 1.67-fold reduction in the percentage of S-phase progenitors in the Plag1 KI/KI cortex (n=3; P<0.01), whereas there was no change in the BrdU labeling index in the Plagl2 KI/KI cortex (Fig. 6D). Similarly, labeling G2/M-phase progenitors with phospho-histone H3 ( pHH3), which marks mitotic cells near the apical surface in E12.5 wild-type (Fig. 6E), Plag1 KI/KI (Fig. 6F) and Plagl2 KI/KI (Fig. 6G) cortices, revealed a threefold reduction in pHH3 + cells in Plag1 mutants (Fig. 6H).
Taken together, these data suggest that Plag1, and not Plagl2, is required to regulate the number of cycling progenitors in S-and G2/M-phases of the cell cycle. We next assessed changes in the expression of progenitor cell markers in E12.5 wild-type (Fig. 6I,M), Plag1 KI/KI (Fig. 6J,N) and Plagl2 KI/KI (Fig. 6K,O) cortices, using Pax6 to label RGCs (Estivill-Torrus et al., 2002) and the T-box transcription factor Tbr2 to label INPs (Arnold et al., 2008;Sessa et al., 2008). Pax6 labeled the same number of RGC progenitors in the VZ in all genotypes (Fig. 6I-L), and Tbr2 labeled the same number of INPs ( Fig. 6M-P). The decrease in VZ proliferation in Plag1 mutants was therefore not translated into an overall change in progenitor cell number, at least by E12.5.
Changes in the BrdU and pHH3-labelling indices in Plag1 mutants might reflect alterations in the ratios between proliferative and differentiating populations rather than changes in proliferation rates. We therefore examined whether Plag1 and/or Plagl2 regulated the differentiation of neocortical progenitors by examining the expression of Tbr1, a marker of early-born, deeplayer neurons (Hevner et al., 2001). There was a small but significant increase in the number of early-born neurons generated in E12.5 Plag1 KI/KI mutants compared to wild type, whereas there were no significant changes in Tbr1 expression in the Plagl2 KI/KI cortices relative to wild type (Fig. 6Q-T). Notably, the differences in proliferation and differentiation observed in Plag1 KI/KI mutants did not translate into alterations in the total number of DAPI + nuclei in the E12.5 neocortex (wild type: 1393±76.4; Plag1 KI/KI : 1188 ±68.8; Plagl2 KI/KI : 1427±80.0; n=3 for each genotype).
Taken together we can conclude that Plag1 is required to maintain the balanced choice between proliferation and differentiation in the E12.5 neocortex, suggesting that Plagl2, which is not mutated (or altered in expression) in Plag1 KI/KI embryos, is not sufficient to rescue this phenotype. In contrast, Plagl2 is not required for the proliferation or differentiation of neocortical progenitors, possibly because Plag1 and altered Zac1 expression levels compensate to some extent for Plagl2 functions.
Plagl2 increases the BrdU labelling index within 24 h post-electroporation Plag1 and Plagl2 are both proto-oncogenes, promoting cell proliferation in a malignant context (Abdollahi, 2007;Van Dyck et al., 2007b). We therefore asked whether the overexpression of these factors in non-transformed, embryonic neural progenitors could similarly promote proliferation, and/or alter differentiation. For this purpose, we used in utero electroporation to introduce Plag1 and/or Plagl2 expression constructs containing an IRES-GFP cassette into dorsal telencephalic progenitors at E12.5, or an empty vector control expressing GFP only. Embryos were harvested 24 h post-electroporation and transfected cells were identified using GFP epifluorescence (Fig. 7A).
In summary, Plagl2 but not Plag1 is sufficient to induce the proliferation of neocortical progenitors, at least within a short 24 h time window.
Plag1 and Plagl2 inhibit neuronal differentiation 72 h post-electroporation, and Plag1 has a delayed effect on increasing the BrdU labelling To assess the effects of Plag1 and Plagl2 on neuronal differentiation and migration, we introduced Plag1 and Plagl2 expression vectors (and pCIG2 control) into E12.5 neocortical progenitors via in utero electroporation, but this time we assessed electroporated brains at E15.5, 72 h post-electroporation. The positions of the GFP + electroporated cells were compared by counting labeled cells in the VZ, SVZ, intermediate zone (IZ) and cortical plate (CP). In control electroporations, most GFP + labeled cells had already reached the CP 72 h post-electroporation (Fig. 8A,D). In contrast, E12.5→E15.5 electroporations of Plag1 (Fig. 8B) and Plagl2 (Fig. 8C) differed, in that more GFP + electroporated cells were concentrated in the SVZ (n=3; P<0.001 for Plag1 and P<0.01 for Plagl2; Fig. 8D) and IZ (n=3; P<0.001 for Plag1 and Plagl2; Fig. 8D), and fewer GFP + cells reached the CP (n=3; P<0.001 for Plag1 and Plagl2; Fig. 8D). Overexpression of either Plag1 or Plagl2 thus strongly perturbs cellular migration, either because overexpressing cells fail to undergo neurogenesis, and/or because these genes impede neuronal migration.
To assess the effects of Plag1 and Plagl2 on neurogenesis in the E12.5 neocortex, at 72 h post-electroporation (at E15.5) we examined the expression of Tbr1 (Fig. 8E-G), which is expressed at high levels in deep-layer neocortical neurons (Englund et al., 2005). In E12.5→E15.5 electroporations the number of Plag1-and Plagl2-transfected cells that expressed Tbr1 was reduced compared to pCIG2 transfections (n=3; P<0.001 for both genes; Fig. 8E-H). These data suggested that Plag1 and Plagl2 do indeed block neuronal differentiation in the neocortex, and the inability of these cells to differentiate likely results in their migratory defects.
The inability of Plagl2-overexpressing progenitors to differentiate was consistent with the increased mitotic activity of these cells 24 h post-electroporation, whereas Plag1 misexpression did not have the same effect after 24 h (Fig. 7E,I). To test whether Plag1 and Plagl2 influenced the proliferative capacity of E12.5 cortical progenitor cells 72 h post-electroporation, we performed a 30 min pulse-label with BrdU ( Fig. 8I-L). In E12.5→E15.5 electroporations there were more GFP + BrdU + S-phase progenitors in Plag1 electroporations compared to pCIG2 control and Plagl2 transfections (n=3; P<0.001; Fig. 8I-L). Plag1-misexpressing cells thus proliferate more and differentiate less, even though they take longer to initiate BrdU uptake compared to Plagl2-overexpressing cells.

DISCUSSION
In this study we investigated the functions of the proto-oncogenes Plag1 and Plagl2 in the developing telencephalon. Our goal was to generate double mutants to assess genetic redundancy, but no live Plag1;Plagl2 double-mutant embryos were collected at E12.5, indicating that these genes can compensate for one another with respect to overall embryonic growth and survival and that at least one gene copy of either Plag1 or Plagl2 is required. Furthermore, we found that the Plagl2 genotype was early-embryonic lethal on a CD1 background, with a reduced number of single mutant embryos obtained at E12.5, restricting our analyses of telencephalic development to this early embryonic stage. Our analysis of E12.5 mutants revealed differences in the functions of Plag1 and Plagl2, including an unexpected and striking patterning defect in the Plagl2 mutant telencephalon, indicating that this gene is required to set the ventral boundary of dorsal gene expression. In contrast, Plag1 is necessary to regulate the balance between proliferation and neuronal differentiation in the developing neocortex.
The Plagl2 mutant patterning defect was somewhat unexpected as we observed an expansion of the ventral pallium without a corresponding decrease in the size of the dLGE, a territory immediately adjacent to the ventral pallium. This phenotype contrasts to defects observed upon the mutation of other patterning genes, including Pax6 and Gsx2, which result in compensatory changes in the sizes of these two domains; in Pax6 mutants, the dLGE expands at the expense of the ventral pallium, whereas the opposite phenotype is observed in Gsx2 mutants (Yun et al., 2001). There is also a repression of a ventral pallial identity upon the deletion of Nr2e1, also called Tlx (tailless), resulting in the loss of Sfrp1 and Dbx1 expression (Stenman et al., 2003a,b). Complex genetic interactions thus regulate the establishment of these two important telencephalic domains, and Plagl2 is an important determinant of ventral pallial domain size. Of note, while we did not observe a reduction in the number of Tbr1 + neurons in the presumptive piriform cortex in Plagl2 mutants, which is derived from the ventral pallium, it may be that analyses at later stages could reveal defects. We were, however, precluded from these later studies by the early embryonic lethality of the Plagl2 mutation.
While we did not observe patterning defects in Plag1 mutants, we did observe that this gene is essential for neocortical progenitor proliferation and to block neuronal differentiation, consistent with its known role as a growth regulator. Interestingly, in gain-of-function studies, Plag1 promotes BrdU uptake and reduces neurogenesis, but the effects on BrdU incorporation were delayed and only revealed 72 h post-electroporation. In contrast, Plagl2 very rapidly induced BrdU uptake in E12.5 neocortical progenitors, within 24 h postelectroporation. While BrdU incorporation is a measure of all progenitors that are in S-phase of the cell cycle, it cannot be used to distinguish between those progenitors that are about to differentiate versus those that will continue to proliferate. However, taken together with the effects on differentiation, our data highly supports the idea that Plag1 and Plagl2 overexpression tips the balance towards more proliferative cell divisions in the early embryonic neocortex.
In summary, despite their related structures and roles as proto-oncogenes, there are key differences in how Plag1 and Plagl2 function in vivo. We discuss the potential reasons for these differences below, and compare Plag1 and Plagl2 functions to Plagl1 (also known as Zac1), the third member of this gene family.
In addition to sharing some common targets, the three transcription factors must also transactivate distinct genes in the developing telencephalon. Some studies have begun to identify transcriptional targets for Plag1 (Voz et al., 2004), Plagl2 (Zheng et al., 2010) and Zac1 (Rraklli et al., 2016;Varrault et al., 2017Varrault et al., , 2006, but which of these genes are important in telencephalic development will require further study. Zac1 misexpression was shown to upregulate the expression of several imprinted genes, consistent with the finding that Zac1 is part of an imprinted gene network (Varrault et al., 2006). Zac1 also induced the expression of negative regulators of the cell cycle, such as the cyclin dependent kinase inhibitors p57 (which is imprinted) and p27, consistent with the ability of Zac1 to promote cell cycle exit (Rraklli et al., 2016). Finally, Zac1 misexpression also induced the ectopic expression of several genes not normally expressed in neural lineages, as well as genes associated with pluripotency, suggesting that one function of Zac1 is to promote a pluripotent state (Rraklli et al., 2016). It will be of interest in the future to see whether Plag1 and Plagl2 also have a similar role in maintaining pluripotency.
We found that Plag1 is required to regulate cell proliferation in the neocortex, which is of particular interest as several genome wide association studies (GWAS) have indicated that a SNP in Plag1 is in one of 27 loci that correlate with height in humans (Yuasa et al., 2010), and additional studies have found correlations with stature or size in various livestock species, including cattle, pigs and horses (Juma et al., 2016). While these are association studies, the underlying assumption is that Plag1 is an important regulator of growth. Interestingly, in our gain-of-function studies, Plag1 was not sufficient to promote proliferation in neocortical progenitors, possibly because it acts in concert with other factors to carry out its growth regulatory role. Igf2 is a downstream transcriptional target of Plag1 that has been implicated in growth control (Voz et al., 2000), and it may be that Plag1 is not sufficient on its own to turn on this transcriptional target in the embryonic neocortex. Another possibility is that Plag1 is sumoylated in the embryonic neocortex, as this post-translational modification has been shown to repress the ability of Plag1 to transactivate downstream targets .
In our study we found that Plagl2 is sufficient but not necessary to promote the proliferation of neocortical progenitor cells. This data is consistent with a previous report showing that misexpression of Plagl2 in p53 −/− adult neural stem cells promotes a proliferative phenotype when these cells are cultured in vitro, with the enriched expression of G1/S cell cycle checkpoint genes (Zheng et al., 2010). Moreover, we found that Plagl2 reduces neuronal differentiation in the embryonic telencephalon, similar to how it functions in glioma cells (Zheng et al., 2010). These results suggest that Plagl2 functions are context-dependent. Indeed, Plagl2 is not oncogenic in all contexts [e.g. pro-apoptotic in response to hypoxia and other cellular stresses  Yang et al. (2009)]. One reason why Plagl2 functions may change in different contexts is that its transcriptional activity is also regulated by post-translational modifications, including sumoylation and acetylation (Guo et al., 2008;Ning et al., 2008;Van Dyck et al., 2004;Zheng and Yang, 2005). Under the right conditions, Plagl2 may promote neocortical progenitor proliferation possibly via its capacity to initiate the transcription of Wnt pathway genes (Zheng et al., 2010), a key proliferative signal in the neocortex (Zheng et al., 2010).
In summary, our study reveals that Plag1 and Plagl2 are not only important regulators of tumorigenesis, but also play important redundant as well as complementary roles in normal CNS development and embryonic survival.

Tissue processing and cryostat sectioning
Embryos were dissected at E12.5 for all loss-of-function studies and at E13.5 for gain-of-function studies. Embryos were fixed overnight at 4°C in 4% paraformaldehyde (PFA) diluted in phosphate buffered saline (PBS) at pH 7.5. To remove the fixative, embryos were washed three times for 10 min in PBS, and then transferred to 20% sucrose/1× PBS overnight. For cyrosectioning, the brains were embedded in optimal cutting temperature (OCT) compound and stored at −80°C. Blocks were then sectioned at 10 µm on a cryostat for immunostaining and RNA in situ hybridization.

BrdU labeling
100 µg/g body weight BrdU (Sigma-Aldrich) was injected intraperitoneally 30 min before the mice were euthanized. Prior to immunolabeling, sections were immersed in 2 N HCl for 15 min at 37°C (Britz et al., 2006). Immunolabeling was then performed as described.

X-Gal staining
Slides were washed with DEPC phosphate buffered saline (PBS) pH7.5 for 5 min, three times. Sections were fixed in fixing solution (0.2% glutaraldehyde, 2% formaldehyde, 5 mM EGTA pH7.3, 2 mM magnesium chloride and 0.1 M sodium phosphate pH7.3 in PBS) for 15 min at room temperature. The slides were then washed with washing solution (0.02% NP40, 2 mM MgCl 2 in PBS) for 10 min, three times. The slides were immersed in prewarmed staining solution [20 mg/ml X-gal-Sigma-Aldrich B4252 dissolved in DMSO, 5 mM K 3 Fe(CN) 6 , 5 mM K 4 Fe(CN) 6 , 2 mM MgCl 2 , 0.02% NP40 in PBS] and incubated in a 37°C water bath for 4 h to overnight protected from light. The tissues were dehydrated in 95% and 100% EtOH, and Xylene at room temperature. After the slides were dry, ∼4 drops of permount was added per slide and mounted with a coverslip for imaging.

In utero electroporation
We performed in utero electroporation as previously described Langevin et al., 2007). Briefly, we introduced 3 μg/μl of a pCIG2 control vector, which expresses GFP, or 3 μg/μl of pCIG2-Plag1 or pCIG2-Plagl2, which express the gene of interest and GFP, into E12.5 telencephalic vesicles using borosilicate needles and a Femtojet microinjector. Using a BTX electroporator, we applied 7 pulses of 55 mV within a 7 s interval to the uterus with the paddles flanking the head of the embryo. The uterus was then put back and embryos underwent normal development until E13.5. pCIG2-Plag1 was generated by PCR amplification of Plag1 from IMAGE clone ID 6328180 using the following primers: Plag1L: AATCTAGAG-ATGGCCACTGTCATTCCTGG; Plag1R: AATCTAGAGGCTACACA-AGCA CCTCGGGT. The amplified Plag1 cDNA was cloned as a blunted XbaI fragment into the blunted EcoRI site of pCIG2. pCIG2-Plagl2 was generated by PCR amplification of Plagl2 from IMAGE clone ID 6405960 using the following primers: Plagl2L: AATCTAGACATGA-CCACATTTTT CACCAG; Plagl2R: AATCTAGACTGAGTTGGGGGA-CCTTCAT. The amplified Plagl2 cDNA was cloned was cloned directionally as an EcoRI fragment into the EcoRI site of pCIG2.

RT-qPCR
We microdissected the dorsal telencephalon from E12.5 embryos as described. RNA was extracted with TRIzol reagent following the instructions of the manufacturer (Thermo Fisher Scientific, Cat#15596026). cDNA was synthesized and RT-qPCR was performed using a RT2 primer assay kit (Qiagen, 330001) according to the manufacturer's instructions. The following RT 2 qPCR primers were obtained from Qiagen: Gapdh (PPM02946E), B2m (PPM03562A), Hrpt (PPM03559F), Plag1 (PPM30678A), Plagl2 (PPM30603B) and Zac1 (PPM03537F). qPCR was performed with cortices from three embryos of each genotype and with three technical replicates for each biological replicate. We used the delta-delta Ct method to calculate relative expression levels, using three housekeeping genes to normalize (Gapdh, B2M, Hrpt).

Imaging, quantitation and statistics
We captured images using OpenLab5 software (Improvision) and a QImaging RETIGA EX digital camera for bright-field images and a Leica DMRXA2 optical microscope for fluorescence imaging. Images were processed in Photoshop CS6 (64 bit; Adobe Systems) and quantification was performed from these images. For quantification we used a minimum of three embryos, and three sections per embryo. Calculation of statistical significance involved a one-way ANOVA with a Tukey correction for multiple comparisons. Graphs and statistical values were generated using GraphPad Prism software.