Cortical GABAergic interneurons in rodents originate from subpallial progenitors and tangentially migrate to the cortex. While the majority of mouse neocortical interneurons are derived from the medial and caudal ganglionic eminence (MGE and CGE, respectively), it remains unknown whether the lateral ganglionic eminence (LGE) also contributes to a subpopulation of cortical interneurons. Here, we show that the transcription factor Sp8 is expressed in one-fifth of adult cortical interneurons, which appear to be derived from both the dorsal LGE and the dorsal CGE (dLGE and dCGE, respectively). Compared with the MGE-derived cortical interneurons, dLGE/dCGE-derived Sp8-expressing (Sp8+) ones are born at later embryonic stages with peak production occurring at embryonic day 15.5. They tangentially migrate mainly along the subventricular/intermediate zone (SVZ/IZ) route; some continue to express mitotic markers (Ki67 and PH3) in the neonatal cortical SVZ/IZ. Sp8+ interneurons continue to radially migrate from the SVZ/IZ into the cortical layers at early postnatal stages. In contrast to MGE-derived interneurons, dLGE/dCGE-derived Sp8+ interneurons follow an outside-in layering pattern, preferentially occupying superficial cortical layers.
The majority of mouse neocortical interneurons are derived from the medial and caudal ganglionic eminence (MGE and CGE, respectively) (Wichterle et al. 2001; Nery et al. 2002; Xu et al. 2004, 2010; Butt et al. 2005; Cobos et al. 2005; Fogarty et al. 2007; Miyoshi et al. 2007). Recently, the analysis of CGE derivatives has been a particularly active area of investigation using Cre fate mapping (Miyoshi et al. 2010) or expression from a 5HT3aR-BACEGFP transgene (Lee et al. 2010; Vucurovic et al. 2010). CGE-derived interneurons do not express parvalbumin (PV) and somatostatin (SOM), recognized markers of MGE-derived interneurons (Wichterle et al. 2001; Nery et al. 2002; Valcanis and Tan 2003; Lopez-Bendito et al. 2004; Xu et al. 2004; Butt et al. 2005; Fogarty et al. 2007). The CGE is continuous with and shares many molecular properties with the lateral ganglionic eminence (LGE) (Flames et al. 2007; Long, Cobos, et al. 2009). Furthermore, there are no definitive molecular markers or specific Cre lines to distinguish between the LGE- and CGE-derived cortical interneurons. Therefore, it is unknown whether the LGE generates a subpopulation of cortical interneurons (Rudy et al. 2011).
Here, we provide evidence that expression of Sp8 transcription factor marks CGE-derived interneurons and perhaps LGE-derived interneurons. Sp8 is a member of the Sp1 zinc finger transcription factor family. Sp8 is widely expressed in the mouse embryonic telencephalon (Bell et al. 2003; Treichel et al. 2003). Its expression in pallial progenitors regulates patterning of the cerebral cortex (Sahara et al. 2007; Zembrzycki et al. 2007), and its subpallial expression regulates differentiation of olfactory bulb (OB) interneurons (Waclaw et al. 2006). Sp8 is strongly expressed in the dorsal LGE (dLGE), a domain that prenatally generates OB interneurons, and it continues to be expressed postnatally in OB interneuron neurogenic regions (Waclaw et al. 2006; Liu et al. 2009; Wei et al. 2011). In the SVZ-RMS-OB system, unlike other transcription factors such as Emx1, Gsh2, Nkx2.1, Dlx2, and Mash1, Sp8 appears to be continuously expressed not only in dividing neuroblasts but also in the majority of mature OB interneurons (Waclaw et al. 2006; Liu et al. 2009). Moreover, Sp8 is required for the production of OB calretinin-expressing (CR+) and PV+ interneurons (Waclaw et al. 2006; Li et al. 2011).
In the present study, we show that Sp8 is expressed in one-fifth of cortical interneurons that are derived from the dorsal CGE (dCGE) but not the MGE. Furthermore, we present evidence that the dLGE may also be a source of Sp8+ cortical interneurons. Like in the SVZ-RMS-OB system, Sp8 is continuously expressed in mitotic migrating neuroblasts in the germinal regions to mature interneurons in the neocortex. Sp8+ interneurons are born later than the MGE-derived ones, and they preferentially occupy superficial cortical layers. During development, most Sp8+ interneurons tangentially migrate through the subventricular/intermediate zone (SVZ/IZ) and only later enter the cortical plate (CP) and marginal zone (MZ). In contrast to the MGE-derived interneurons that are generated in an inside-out manner, dLGE/dCGE-derived Sp8+ interneurons follow an outside-in layering pattern.
Materials and Methods
Adult CD1 mice were purchased from Shanghai SLAC Laboratory Animal Co. Ltd. Dlx5/6-cre-IRES-EGFP (Dlx5/6-CIE) mice were the generous gift of Dr Kenneth Campbell (Stenman et al. 2003). Nkx2.1-cre (Xu et al. 2008), Nestin-cre (Tronche et al. 1999), Rosa-YFP (Srinivas et al. 2001), and Z/EG (Novak et al. 2000) mice were obtained from the Jackson Laboratory (Bar Harbor, ME). All lines were maintained in a mixed genetic background of C57BL/6J and CD1. The day of vaginal plug detection was designated as embryonic day 0.5 (E0.5), and the day of birth was designated as postnatal day 0 (P0). All experiments were performed in accordance with institutional guidelines.
For bromodeoxyuridine (BrdU) birth dating, timed pregnant female mice (E10.5, E11.5, E12.5, E13.5, E14.5, E15.5, E16.5, E17.5, and E18.5) and P0 mice received a single intraperitoneal injection of BrdU (50 mg/kg body weight, Sigma). For labeling dividing cells at E13.5 and P0, mice were sacrificed 2 h after BrdU injection. Other mice were sacrificed at either P0 or P28.
Postnatal brains were fixed by intracardiac perfusion with 0.9% saline followed by 4% paraformaldehyde (PFA) and then postfixed with 4% PFA overnight at 4 °C. Embryonic brains were dissected and fixed overnight in 4% (PFA) at 4 °C. Brains were cryoprotected for at least 24 h in 30% sucrose in 0.1 M phosphate buffer and then frozen in embedding medium (O.C.T., Sakura Finetek) on a dry ice/ethanol slush. Immunohistochemistry staining was performed on 30 μm free-floating sections for postnatal brain samples and 50 μm for embryonic brain samples. Primary antibodies were incubated overnight. For BrdU immunostaining, sections were pretreated with 2 N HCl for 1 h at room temperature (RT).
For primary antibodies, we used rat anti-BrdU (1/400, OBT0030S; Accurate Chemical and Scientific Corporation), rabbit anti-calretinin (CR) (1/3000, AB5054; Chemicon), rabbit anti-GFAP (1/500, Z0334; Dako), chicken anti-GFP (1/800, GFP-1020; Aves labs), rabbit anti-Ki67 (1/400, VP-K451; Vector Laboratories), mouse anti-NeuN (1/400, mab377; Chemicon), rabbit anti-NPY (1:500, 22940; Incstar), mouse anti-parvalbumin (1/400, MAB1572; Chemicon), rabbit anti-parvalbumin (1/1000, PV25; Swant), rabbit anti-PH3 (1/400, H0412; Sigma), mouse anti-Reelin (CR-50; 1/200, D223-3; MBL), rabbit anti-somatostatin (1/100, sc-13099; Santa Cruz), rabbit anti-VIP (1:500, 20077; Incstar), rabbit-anti-Cux1 (CDP, M222; 1/1000, sc-13024; Santa Cruz), mouse anti-COUP-TFII (1/300, PP-H7147-00; Perseus Proteomics), rabbit anti-Isl1 (1/300, ab20670; Abcam), rabbit anti-Nkx2-1 (1/300, sc-13040; Santa Cruz), rabbit anti-Olig2 (1/500, AB9610; Chemicon), rabbit anti-Pax6 (1/300, PD022; MBL), rabbit anti-Sox6 (1:4000, ab30455; Abcam), goat anti-Sp8 (1/500, sc-104661; Santa Cruz), and rabbit anti-Tbr1(1/200, ab31940; Abcam). Secondary antibodies against the appropriate species were incubated for 2 h at RT (all from The Jackson Laboratory, 1:200). Omission of primary antibodies eliminated staining. All sections were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) (Sigma, 400 ng/mL, 5 min).
Confocal Imaging and Cell Counting
All images presented in the study were acquired using the Olympus FV1000 confocal microscope system. Images were cropped, adjusted, and optimized in Photoshop CS2. Cell counting was performed under an Olympus BX 51 microscope.
For quantification of cortical interneurons from P28 mice, 6 coronal sections (n = 3 mice) spanning primary somatosensory cortex were selected. Cortical layers were delineated by DAPI-stained nucleus.
For quantification of interneurons from embryonic and P0 mice, 6–8 anatomically matched sections spanning the rostral–caudal extent of the telencephalon (n = 3) were selected. Cell counting was performed in the cortex adjacent to the pallial–subpallial boundary.
For quantification of Ki67+/Sp8+ and BrdU+/Sp8+ cells in the dLGE/dCGE of E13.5 mice, 6 anatomically matched sections spanning the rostral–caudal extent of the telencephalon (n = 3) were quantified.
For quantification of Sp8+/BrdU+ interneurons at P0 mice, 6 anatomically matched sections spanning the putative primary somatosensory cortex (n = 3) were selected. All data are presented as the means ± standard error of the mean.
Sp8 Is a Novel Marker for Specific Cortical Interneuron Subtypes
Using in situ hybridization and immunohistochemistry, previous studies have shown that Sp8 messenger RNA and its protein are expressed in the subpallial germinal regions at E13.5 and E15.5 (Waclaw et al. 2006; Long, Cobos, et al. 2009; Long, Swan, et al. 2009). While these Sp8+ progenitors become OB interneurons (Waclaw et al. 2006), it is unknown whether Sp8+ cells also become other types of interneurons. To investigate this, we performed Sp8/green fluorescent protein (GFP) double immunostaining on brain sections of Dlx5/6-CIE; Z/EG mice at P28, in which GFP labeled nearly all neocortical interneurons. The Sp8 antibody utilized in this study is specific in immunohistochemistry, as no Sp8 immunoreactivity was detectable in the OB section of Sp8 conditional knockout (Dlx5/6-CIE; Sp8flox/flox) mice (Li et al. 2011). We quantified the double labeling in the primary somatosensory cortex and found that virtually all (96 ± 1%) Sp8+ cells were GFP labeled, whereas 20% GFP+ cells were Sp8+ (Fig. 1A,B,G). Further investigation showed that Sp8 was expressed by a subpopulation of Reelin+, VIP+, NPY+, and bipolar CR+ cortical interneurons (Fig. 1C–F). By contrast, Sp8 was not expressed in GFAP+, Olig2+, Cux1+, or Tbr1+ cortical cells (data not shown), indicating that most glial cells and cortical projection neurons do not express Sp8. Sp8+ cells preferentially occupy superficial cortical layers (Fig. 1H,I). In Dlx5/6-CIE; Z/EG mice, more than 70% Sp8+/GFP+ interneurons were located within layers I and II/III (28 ± 1% and 45 ± 1%, respectively; Fig. 1H), and a large population of interneurons (GFP+ cells) in layers I and II/III express Sp8 (84 ± 1% and 39 ± 2%, respectively; Fig. 1H). Furthermore, the majority of Sp8+/Reelin+ interneurons (83 ± 3%) were located within layer I, accounting for most (89 ± 2%) Sp8+ interneurons in layer I, whereas Sp8+/VIP+ and Sp8+/NPY+ interneurons were located mainly within layer II/III (61 ± 3% and 58 ± 1%, respectively; Fig. 1H,I). Because very rare Sp8+/Reelin+/VIP+ interneurons were found in the cortex, Sp8+/Reelin+ and Sp8+/VIP+ interneurons represented 2 nonoverlapping subpopulations of Sp8+ interneurons, which account for nine-tenths (87 ± 2%) of Sp8+ interneurons.
Sp8+ Cortical Interneurons Are Not Derived from the MGE or Preoptic Area
PV+ and SOM+ cortical interneurons are derived from the MGE (Wichterle et al. 2001; Nery et al. 2002; Valcanis and Tan 2003; Lopez-Bendito et al. 2004; Xu et al. 2004; Butt et al. 2005; Fogarty et al. 2007). However, we found no evidence of Sp8+/PV+ or Sp8+/SOM+ cells in the cortex (data not shown). Sox6, a marker for MGE-derived cortical interneurons (Azim et al. 2009; Batista-Brito et al. 2009), is also not expressed by the Sp8+ interneurons (Fig. 2A). This suggests that Sp8+ cortical interneurons are not derived from the MGE. No Sp8+/YFP+ cells exist in the cortex of Nkx2.1-cre; Rosa-YFP mice (Fig. 2B), further supporting this conclusion. Moreover, this result also excludes the possibility that Sp8+ cortical interneurons are derived from the preoptic area (POA) because Nkx2.1 is also expressed in this region (Gelman et al. 2009; Flandin et al. 2010).
Previous studies have shown that the transcription factor COUP-TFII is preferentially expressed by CGE-derived cortical interneurons, although it also labels a subpopulation of MGE-derived cortical interneurons (Kanatani et al. 2008; Willi-Monnerat et al. 2008). In the primary somatosensory cortex at P28, we found that one-third (34 ± 3%) of Sp8+ interneurons were COUP-TFII+; conversely, one-third (38 ± 4%) of COUP-TFII+ interneurons were Sp8+. Additionally, one-third (31 ± 1%) of COUP-TFII+ interneurons expressed Sox6, providing evidence that these interneurons originate from the MGE. In Dlx5/6-CIE; Z/EG mice cortex, Sp8+/COUP-TFII+/GFP+ interneurons are mainly found in the superficial layers. Indeed, in layer I, the vast majority (91 ± 2%) of Sp8+ interneurons colocalize with COUP-TFII and vice versa (87 ± 2%) (Fig. 2C–F,H). By contrast, the MGE-derived Sox6+/COUP-TFII+/GFP+ interneurons are mainly located in the deep layers (Fig. 2C–F,H). The above observation indicates that COUP-TFII+ cells in the CGE give rise to one-third of Sp8+ interneurons in the cortex. Importantly, we found that Sp8+, Sox6+, and COUP-TFII+ cells account for nearly 90% of cortical interneurons (Fig. 2G).
Sp8+ Cortical Interneurons Appear to Originate from the dCGE and dLGE and Largely Tangentially Migrate through the SVZ/IZ
To examine the expression pattern of Sp8 in the telencephalon, we systematically analyzed the distribution of Sp8+ cells at E13.5. Sp8 is expressed in the pallial ventricular zone (VZ) in a high to low rostrodorsal to caudoventral gradient (Fig. 3A–D), consistent with previous studies (Sahara et al. 2007). In the dLGE and dCGE, Sp8 is strongly expressed in most SVZ cells, with rare Sp8+ cells in the VZ (Fig. 3). Interestingly, we found Sp8+ cells emerging from the dLGE and dCGE and appearing to migrate into the cortex (Fig. 3A1–D1). These Sp8+ cells are likely to be immature interneurons that are not derived from the MGE as the majority of them express GFP but not Sox6 in Dlx5/6-CIE; Z/EG mice (Fig. 3A1–D1).
In the E13.5 LGE, Sp8 is not expressed in Isl1+ cells, providing further evidence that the Sp8+ cells are restricted to the dLGE (Fig. 3E,F). In the dCGE, Sp8 extensively overlaps with COUP-TFII (Fig. 3F,G,H). In the ventral CGE (vCGE), although a small number of Sp8+/COUP-TFII+ cells are also found, the majority of COUP-TFII+ cells do not express Sp8 (Fig. 3F,G,H). Sp8/Pax6 double immunostaining revealed that Sp8+ cells in the E13.5 dLGE do not extend across the pallial–subpallial boundary (Fig. 3I; Waclaw et al. 2006). Interestingly, in the lateral cortical stream (Carney et al. 2006), many Pax6+ cells that migrate into the piriform cortex also express Sp8 (Fig. 3I), but no Sp8+/Pax6+ cells were observed in the dorsal cortex.
Proliferation analysis in the dLGE and dCGE at E13.5 showed that a subpopulation of Sp8+ cells in the SVZ coexpress Ki67 (Fig. 3J,J1; 34 ± 2%, Ki67 and Sp8 double+ cells/Sp8+ cells, n = 3 mice). Thus, a subpopulation of mitotically active cells expresses Sp8. To confirm this conclusion, E13.5 mice were killed 2 h after a single BrdU injection. Sp8/BrdU double immunostaining revealed Sp8+/BrdU+ cells in the SVZ of the dLGE and dCGE (Fig. 3K,K1; 19 ± 3%, BrdU and Sp8 double+ cells/Sp8+ cells, n = 3 mice). Taken together, these results provide evidence that Sp8 is expressed in a subpopulation of neocortical interneurons, from their neuroblast stage to adulthood, similar to its expression in the SVZ-RMS-OB system.
Many Sp8+ cells in the dLGE contribute to OB interneurons (Waclaw et al. 2006). It is possible that Sp8+ cells in the dLGE also contribute to Sp8+ cortical interneurons. To further investigate this, we again took advantage of Dlx5/6-CIE; Z/EG mice to label tangentially migrating interneurons at early embryonic stages. A migratory stream of Sp8+ cells entering the cortex from the dLGE was found as early as E12.5 (Fig. 4A,B) (20 Sp8+ cells/section, totally 18 sections were analyzed, n = 3 mice). At this point in time, however, fewer Sp8+ cells migrate out of the dCGE (Fig. 4C,D) (2.4 Sp8+ cells/section, totally 18 sections were analyzed, n = 3 mice), providing evidence that the dLGE could be a source for cortical interneurons. We also found that tangential migration behavior of Sp8+ cortical interneurons is unique. MGE-derived cortical interneurons migrate into the cortex via SVZ/IZ and MZ streams (Lavdas et al. 1999). Indeed, at E15.5, there was a large number of MGE-derived cortical interneurons in the SVZ/IZ, CP/subplate (SP), and MZ (Fig. 4F,H). By contrast, Sp8+ cells were primarily in the SVZ/IZ; rare Sp8+ cells were in the MZ and CP/SP at E15.5 (Fig. 4E–H). By E17.5, increasing numbers of Sp8+ interneurons were observed in the MZ and CP/SP (Fig. 4I–L). Quantificating the number of Sp8+ cells in the SVZ/IZ, CP/SP, and MZ at E13.5, E15.5, E17.5, and P0 demonstrated that there were always more Sp8+ cells in the SVZ/IZ than the MZ (Fig. 4M). Moreover, from E13.5 to P0, there was an increasing proportion of Sp8+/GFP+ cells in the SVZ/IZ of Dlx5/6-CIE; Z/EG mice (Fig. 4N). Taken together, these data suggest that Sp8+ interneurons migrate mainly within the deeper SVZ/IZ route and enter the cortex roughly 2 days later than the MGE-derived ones. This is consistent with previous observations on the migration of LGE/CGE-derived interneurons (Rubin et al. 2010).
Radial Migration and Outside-In Layering of Sp8+ Cortical Interneurons
While many Sp8+ interneurons were observed in the cortical layers at P0, there were still a large number of Sp8+ cells accumulating in the cortical SVZ and IZ (Fig. 5A–A3). More surprisingly, radial migration of Sp8+ interneurons from the SVZ/IZ or SVZ/CC (corpus callosum, CC) into the cortical layers from P0 to P7 was observed (Fig. 5); they frequently formed elongated chain-like structures similar to those observed in the adult rostral migratory system (RMS) (Lois et al. 1996; Inta et al. 2008). Furthermore, many of them had a leading process oriented toward the cortical layers (Fig. 5A1–A3). At P0, most GFP+ interneurons (66 ± 4%) in the IZ of Dlx5/6-CIE; Z/EG mice expressed Sp8. By contrast, only a small number of GFP+ interneurons (26 ± 3%) in the IZ expressed Sox6, as most Sox6+ interneurons were already distributed in the cortical layers (Fig. 5A–A3). This further supports the notion that Sp8+ interneurons enter the cortical layers later than the MGE-derived population.
Previous studies suggested that few neocortical interneurons are generated within the proliferative zone of neocortex (Anderson et al. 2001, 2002). More recently, one study demonstrated that some dividing cortical interneuron progenitors exist in the pallial SVZ/IZ or SVZ/CC; they originate from the LGE or MGE (Wu et al. 2011). Consistent with this report, in P0 Dlx5/6-CIE, Z/EG mice, we found Sp8+/Ki67+/GFP+ and Sp8+/PH3+/GFP+ cells in the pallial IZ (Fig. 5E–G4), suggesting that they are mitotically active. This conclusion is further confirmed by BrdU pulse labeling (Fig. 5H–H2). Because the MGE does not contribute to Sp8+ interneurons, these dividing Sp8+ cells in the IZ are likely derived from the LGE and/or CGE.
Using traditional BrdU birth-dating analysis, we identified that the majority of Sp8+ cortical interneurons were born between E13.5 and E17.5 with a peak at E15.5 (Fig. 6A–C,G,I), indicating that the majority of Sp8+ cortical interneurons were generated later than the MGE-derived population (Miyoshi et al. 2007, 2010).
Sp8+ cortical interneurons follow an outside-in layering pattern (Fig. 6A–C,H). Sp8+ interneurons born at E13.5–E15.5 are mainly distributed in the upper layers, whereas those born at later embryonic stages are preferentially distributed in the deeper layers (Fig. 6A–C,H,I). Consistent with the existence of dividing interneuron progenitors in the pallial SVZ/IZ or SVZ/CC, we found that about 1% of Sp8+ cortical interneurons that located in deep layer V and VI were generated at P0 (Fig. 6G,H).
As shown above, there were many Sp8+ cells accumulating in the cortical SVZ/IZ (Fig. 5A). E13.5, E15.5, or E17.5 mice received a single injection of BrdU. These mice were sacrificed at P0, and the numbers of BrdU+/Sp8+ cells in the cortical IZ, CP/SP, and MZ were quantified. We observed that many Sp8+ interneurons born at E13.5, E15.5, and E17.5 were in the P0 pallial SVZ/IZ (Fig. 6D–F,J), suggesting that at least some Sp8+ cells accumulate in the SVZ/IZ before they migrated into the cortical layers. Accordingly, from E13.5 to E17.5, the increase in the proportion of BrdU+/Sp8+ cells in the IZ was found, whereas there was a decrease in the proportion of them in the MZ and CP/SP (Fig. 6D–F,J). Taken together, our analysis of BrdU birth dating demonstrated that Sp8+ interneurons born at early embryonic stages preferentially occupy the upper layers, while late-born ones preferentially occupy the deeper layers, resulting in outside-in layering of the Sp8+ cortical interneurons.
We provide novel evidence that the Sp8 transcription factor is continuously expressed in dCGE-derived mouse neocortical interneurons from the progenitor to mature interneuron state. Furthermore, the dLGE may also be a source for these interneurons. In the adult neocortex, Sp8+ interneurons account for one-fifth of all interneurons. Most Sp8+ interneurons are born later than most MGE-derived interneurons, with peak production at E15.5. Sp8+ interneurons tangentially migrate into the cortex mainly though the SVZ/IZ route and then are radially arrayed in an outside-in pattern, preferentially occupying superficial cortical layers.
Classification of Mouse Cortical Interneurons Using 3 Transcription Factors
The cerebral cortex consists of 2 main classes of neurons, pyramidal projection neurons and GABAergic interneurons. Interneurons comprise 20–30% of the cortical neurons and play an important role in the function of the cortex. During the past 2 decades, regional origins and genetic diversity of cortical interneurons have been extensively investigated (Marin and Rubenstein 2001; Wonders and Anderson 2006; Fishell and Rudy 2011; Gelman et al. 2012). Calcium-binding proteins, neuropeptides, and neurotransmitters are typically used for classification of mature cortical interneurons. For example, PV and SOM are expressed in MGE-derived cortical interneurons (Wichterle et al. 2001; Nery et al. 2002; Valcanis and Tan 2003; Lopez-Bendito et al. 2004; Xu et al. 2004; Butt et al. 2005; Fogarty et al. 2007). Transcription factors have also been used for the classification of cortical interneurons, such as Lhx6 and Sox6, which are widely employed for identification of the MGE-derived cortical interneurons (Cobos et al. 2005; Du et al. 2008; Azim et al. 2009; Batista-Brito et al. 2009). However, because of the lack of specific transcription factors to mark those mature cortical interneurons that are generated outside the MGE, classification of cortical interneurons using transcription factors has been hampered. In the present study, we found that Sp8 specifically labels a subpopulation of neocortical interneurons that are not derived from the MGE. Furthermore, Sp8 is continuously expressed by dividing neuroblasts in the germinal regions and mature neocortical interneurons. This expression pattern enables us to identify embryonic origins and migratory paths of Sp8+ interneurons. However, it is worth noting that Sp8 does not label all interneurons generated outside the MGE. Indeed, we found that many COUP-TFII+ cells in the neocortex do not express Sp8 or Sox6. These cells are most likely derived from the vCGE, as there are a large number of COUP-TFII+ cells that do not express Sp8 or Sox6 in this region. We also note that Sp8+, Sox6+, and COUP-TFII+ cells account for nearly 90% of cortical interneurons. The other 10% of cortical interneurons remains to be identified. However, we cannot exclude the possibility that Sox6, Sp8, or COUP-TFII is downregulated in some mature cortical interneurons.
The dLGE May Contribute to a Subpopulation of Neocortical Interneurons
It is recognized that cortical interneurons are derived from the MGE (Wichterle et al. 2001; Nery et al. 2002; Valcanis and Tan 2003; Lopez-Bendito et al. 2004; Xu et al. 2004; Butt et al. 2005; Fogarty et al. 2007), CGE (Nery et al. 2002; Xu et al. 2004), and POA (Gelman et al. 2009). Previous studies have also suggested that the LGE contributes to a small number of interneurons in the neocortex (Anderson et al. 2001; Jimenez et al. 2002). However, as there are no CGE- or LGE-specific molecular markers or specific Cre driver lines, it is still uncertain whether the LGE generates cortical interneurons (Rudy et al. 2011).
Based on our current observations and those of other groups, it is possible that some Sp8+ progenitors in the dLGE generate cortical interneurons. This speculation is based on several findings:
1) There is a migratory stream of Sp8+ cells from the dLGE to the cortex from E12.5 to P0; at E12.5, the dLGE stream is larger than the dCGE stream. Using a subtractive fate-mapping approach to specifically labeled LGE/CGE-derived interneurons, previous studies observed a similar migratory stream (Rubin et al. 2010).
2) One-third of Sp8+ cortical interneurons were COUP-TFII+; COUP-TFII is preferentially expressed by CGE-derived cortical interneurons. (Kanatani et al. 2008; Willi-Monnerat et al. 2008). The majority of Sp8+/COUP-TFII+ cells are found in the dCGE. This suggests that the CGE gives rise to at least one-third of Sp8+ cortical interneurons. Thus, two-thirds of Sp8+ cortical interneurons do not express COUP-TFII. Since we found a small number of Sp8+ cells in the CGE that do not express COUP-TFII, this population of Sp8+ cells may also contribute to a small number of Sp8+ cortical interneurons. Sp8 is also strongly expressed in the septum (Fig. 3B,C; Waclaw et al. 2006), but the septum does not generate cortical interneurons (Rubin et al. 2010). Because COUP-TFII is largely excluded from the LGE (Kanatani et al. 2008; Willi-Monnerat et al. 2008), we propose that the dLGE generates the Sp8+/COUP-TFII immunonegative cortical interneurons, which would constitute at least 50% of Sp8+ cortical interneurons (10% of total cortical interneurons).
Using Mash1-BACCreER mice and 5HT3aR-BACEGFP mice, recent studies suggest that the CGE produces a large and diverse population of cortical interneurons (Lee et al. 2010; Miyoshi et al. 2010). Interestingly, the characterizations of GFP-labeled cortical interneurons in these mice are very similar to Sp8+ cortical interneurons described in this study (Lee et al. 2010; Miyoshi et al. 2010; Vucurovic et al. 2010). Administrating tamoxifen to Mash1-BACCreER; RCE:loxP mice, both the LGE- and CGE-derived interneurons but not MGE-derived interneurons were labeled (Miyoshi et al. 2010). In 5HT3aR-BACEGFP mice, it is worth noting that GFP also labels both the LGE- and the CGE-derived interneurons because numerous OB interneurons are GFP+ (Inta et al. 2008). Thus, these studies do not exclude the possibility that the LGE may also contribute a subpopulation of cortical interneurons. Similarly, although we observed some Sp8+ cells that migrate from the dLGE into the cortex as early as E12.5, we also cannot exclude the possibility that these Sp8+ cells are the CGE-derived cells that migrate through the LGE into the cortex (Wichterle et al. 2001).
Outside-In Layering of Sp8+ Cortical Interneurons
Like cortical projection neurons, the MGE-derived neocortical interneurons (PV+ and SOM+) also follow an inside-out layering pattern in the neocortex (Miyoshi et al. 2007). Our BrdU birth-dating analysis, however, reveals that Sp8+ interneurons born at early embryonic stages preferentially migrate into cortical layers earlier and eventually occupy the upper layers, while late-born ones preferentially occupy the deeper layers, which results in outside-in layering of the Sp8+ cortical interneurons. Previous studies on rat neocortex have also shown that while PV+ interneurons follow the inside-out neurogenesis pattern, CR+ interneurons show an outside-in gradient (Rymar and Sadikot 2007). In the present study, we show that the majority of Sp8+ interneurons occupy superficial cortical layers, consistent with their CGE origin (Lee et al. 2010; Miyoshi et al. 2010; Vucurovic et al. 2010). The distinct laminar distributions of Sp8+- (CGE/LGE) and Sp8− (MGE)-derived interneurons may partially reflect the fact that the majority of Sp8+ interneurons are generated later than MGE-derived interneurons. However, as there may be distinct molecular cues regulating laminar positioning of interneurons (Lodato et al. 2011), these laminar differences may reflect their distinct molecular affinities for these layers.
The tangential migration behavior of Sp8+ cortical interneurons is distinct; while MGE-derived cortical interneurons migrate into the cortex via both the SVZ/IZ and MZ streams (Lavdas et al. 1999; Sussel et al. 1999), the majority of Sp8+ interneurons enter the cortex mainly within the deeper SVZ/IZ route. We only found a relatively small number of Sp8+ interneurons in the MZ at the late embryonic stage. During the first postnatal week, we found many immature interneurons radially migrating from the SVZ/IZ into the CP/cortical layers, consistent with the previous findings (Hevner et al. 2004; Inta et al. 2008). Moreover, some cortical interneurons may be generated from Sp8+/Ki67+ and Sp8+/PH3+ cells that we observe at P0; these progenitors appear to migrate tangentially from the LGE/CGE into the pallial SVZ/IZ or SVZ/CC.
The Function of Sp8 in the Development of Cortical Interneurons Remains to Be Elucidated
In the mouse embryonic telencephalon, Sp8 expression in pallial progenitors regulates patterning of cortical areas (Sahara et al. 2007; Zembrzycki et al. 2007). Sp8’s function in the ganglionic eminences has been studied in conditional mutants, demonstrating that it is required for the specification, migration, and survival of CR+ interneurons in the OB from late embryonic to postnatal stages (Waclaw et al. 2006). More recently, we observed a severe reduction in PV+ interneurons in the external plexiform layer of the OB in Sp8 conditional mutant mice (Dlx5/6-CIE; Sp8flox/flox mice) (Li et al. 2011). To investigate the function of Sp8 in neocortical interneurons, we analyzed the primary somatosensory cortex of Dlx5/6-CIE; Sp8flox/flox and Dlx5/6-CIE; Sp8flox/flox; Z/EG mice at P28, P42, and P90. Compared with wild-type mice, loss of Sp8 function did not alter the overall distribution of neocortical interneuron in Dlx5/6-CIE; Sp8flox/flox; Z/EG mice indicated by GFP immunostaining (data not shown). There were also no difference in the absolute numbers of Reelin+ and COUP-TFII+ interneurons in layer I, VIP+ cells in all layers, or the laminar distribution of Reelin+, VIP+, NPY+, CR+, and COUP-TFII+ interneurons in the neocortex (data not shown).
To delete Sp8 from a subset of primary neural stem cells (neuroepithelium), we also used the transgenic mouse nestin-cre (Tronche et al. 1999; Graus-Porta et al. 2001), which exhibits excision of the floxed alleles in neuroepithelium at E10.5 based on loss of Sp8 immunostaining (cortical patterning is almost completed at this time point). Nestin-Cre; Sp8flox/flox mouse brains were indistinguishable from their wild-type and Nestin-Cre; Sp8flox/+ littermates at P28, P42, and P90 except that the OB was grossly reduced in the size. Again, no phenotypic abnormalities in neocortical interneurons were observed. It is worth noting that Foxg1-Cre–mediated conditional Sp8 mutant brains exhibit multiple malformations, including the dorsoventral patterning defects at the medial telencephalic wall and a variable dysgenesis of the midline (Zembrzycki et al. 2007), showing that Sp8 has an essential role in patterning of the embryonic forebrain (Bell et al. 2003; Treichel et al. 2003; Sahara et al. 2007; Zembrzycki et al. 2007). However, currently, it is unclear what roles Sp8 plays in the development of a subpopulation of neocortical interneurons. Given that transcription factor Sp9 shares significant sequence homology with Sp8 (Kawakami et al. 2004) and also exhibits a overlapping patterns of expression in the subpallial germinal regions of embryonic telencephalon (Long, Cobos, et al. 2009; Long, Swan, et al. 2009), Sp9’s function is likely to compensate in the absence of Sp8.
National Basic Research Program of China (grant number 2010CB945500 and 2011CB504400); National Natural Science Foundation of China (grant 30900425, 30970949, 30990261, and 30821002) to Z.Y.
We thank Dr Kenneth Campbell for sending the Dlx5/6-cre-IRES-EGFP mice. We would also like to thank members of the Yang laboratory for helpful comments on the manuscript. Conflict of Interest : None declared.