A restricted period for formation of outer subventricular zone defined by Cdh1 and Trnp1 levels

The outer subventricular zone (OSVZ) is a germinal layer playing key roles in the development of the neocortex, with particular relevance in gyrencephalic species such as human and ferret, where it contains abundant basal radial glia cells (bRGCs) that promote cortical expansion. Here we identify a brief period in ferret embryonic development when apical RGCs generate a burst of bRGCs that become founders of the OSVZ. After this period, bRGCs in the OSVZ proliferate and self-renew exclusively locally, thereby forming a self-sustained lineage independent from the other germinal layers. The time window for the brief period of OSVZ bRGC production is delineated by the coincident downregulation of Cdh1 and Trnp1, and their upregulation reduces bRGC production and prevents OSVZ seeding. This mechanism in cortical development may have key relevance in brain evolution and disease.

bRGCs in ISVZ at P6 labeled after ventricular injection of rv::Gfp at P1, displaying 10 PhVim label without or with a basal process, and relative abundance (n = 54 cells, 4 11 animals; mean value). (f-j) PhVim+ cells in VZ,ISVZ and OSVZ at P6 and E38,12 without or with a basal process, expressing Pax6 or Tbr2, and relative abundance (j)        Our marker expression analyses demonstrated that the vast majority of GFP+ cells 140 morphologically identified as aRGCs and bRGCs were positive for Ki67 and Pax6, and in 141 addition that 23.7% of aRGCs and 34.3% of bRGCs also expressed Tbr2 (Fig. 1). This was 142 surprising because in mouse Tbr2 is a marker of IPCs but not RGCs 2-4 , also reported 143 previously in ferret and human 5-8 . However, these observations were fully consistent with 144 more recent analyses in macaque embryos 9 . To confirm these findings we performed anti-  Fig. 1c-e). To reconcile these results  Fig. 1f,g,j). Similarly, Tbr2 was 156 expressed at similar frequencies by cells with and without a PhVim+ basal process in all three 157 germinal layers (Supplementary Fig. 1h-j). This was contrary to mouse, where bRGCs are including cells in the astrocyte and oligodendrocyte lineages (Fig. 1, Supplementary Fig.   171 1a,b). All these cell types were revealed by GFP regardless of the layer of rv::Gfp delivery, 172 progressively accumulated in IZ and CP at late stages (Supplementary Fig. 2a,b), and 173 systematically predominated in the OSVZ and IZ/CP, particularly MNs and DNs 174 (Supplementary Fig. 2c). As expected, DNs were only found in the upper CP and increased 175 in abundance at later stages (P14; Supplementary Fig. 2c).

176
MP cells had the typical morphology of IPCs 16,17 , but their relative abundance was 177 very low (<7% of GFP+ cells; Supplementary Fig. 2c,d), especially considering that many 178 cells in GFP+ clones eventually differentiated as neurons (Supplementary Fig. 2c).

179
Moreover, SVZ cells with multipolar morphology include IPCs and newborn neurons 18 . To 180 specifically identify IPCs we stained against Ki67 (marker of cycling cells) and Tbr2 (marker 181 of IPCs and newborn neurons). This analysis was performed at P6 after rv::Gfp injection in 182 VZ at P1, to obtain the largest number of GFP+ MP cells (Supplementary Fig. 2, 3a).

183
Whereas 100% of MPs were Tbr2+, only 14.3% were Ki67+, indicating that only a minority 184 of the already few MPs were IPCs (Supplementary Fig. 3b,d). In contrast to the unexpected 185 scarcity of MP cells, MN cells were very abundant (Supplementary Fig. 2), so we analyzed 186 if these could be IPCs with bipolar morphology. Only 25.0% of MNs were Tbr2+ and none were Ki67+ (Supplementary Fig. 3c,d), consistent with these cells not being IPCs but rather 188 newborn cortical neurons.

189
The above results using rv::Gfp seemed to indicate that IPCs were extremely rare in 190 ferret, contrary to mouse 2,3,19 . An alternative was that our pleiotropic retroviral vectors might 191 somehow have a different cellular specificity between mouse and ferret. To discard this 192 possibility we injected rv::Gfp in mouse embryos at E14.5 and analyzed at E16.5, a period of 193 cortical development equivalent to P1-P6 in ferret (Supplementary Fig. 3e). MP cells 194 represented 29.7% of all GFP+ cells in mouse, eight times more than in ferret (3.6%; 195 Supplementary Fig. 3f-h). Ki67 stains revealed that in mouse embryos 12.4% of all GFP+ 196 cells were putative IPCs (Ki67+ MP cells; Supplementary Fig. 3g), while these represented 197 only 0.5% in ferret (Supplementary Fig. 3h). This further supported that IPCs may be very Ki67 was again extremely low or null (Supplementary Fig. 3o IPCs are also very scarce and bRGCs are highly neurogenic 2,9,16,17 .

218
Comparison between DEGS across the critical period and upon Trnp1 knock-down 219 Our microarray gene expression analysis revealed the existence of two sets of differentially-220 expressed genes (DEGs): "Early-change" genes, whose expression levels changed 221 significantly only between E30 and E34; and "Late-change" genes, whose expression changed 222 only between E34 and P1 (Fig. 5). This distinguished those genes that changed early with the 223 increase in bRGC generation, and those that changed late with the decrease of bRGC 224 generation. Among the "late" genes we found Trnp1, a gene encoding for a DNA-associated 225 protein which has been suggested to regulate the expression or function of other genes, and 226 previously shown to block bRGC production in the embryonic mouse cortex 1 . We compared  Fig. 5a). Because similar biological functions may be performed by 231 redundant genes, we next compared the functional gene annotation term clusters between 232 these three sets of DEGs (Supplementary Fig. 5b). With this analysis we found some 233 coherent similarities, such as with the term "organelle": top 3 cluster in down-regulated late 234 genes and top 4 cluster in up-regulated Trnp1-KD genes; and also the related terms 235 "ribonucleotide binding" and "regulation of transcription", top 2 cluster in down-regulated 236 late genes and top 1 cluster in up-regulated Trnp1-KD genes, respectively.