Transferring Mouse Emx1 and Emx2 Lentiviruses into the Chicken Embryonic Brain and Their Implication to the Organization and Evolution of the Amniote Pallium

The development and evolution of the brain are a fascinating research topic [1]. Homeobox genes are highly conserved during the evolution of animals as suggested by the ability of mouse Emx2 to rescue the head phenotype in fly ems mutants [2]. Among vertebrates, Pax6 is expressed throughout the pallium, and Emx1 and Emx2 are expressed in the dorsal pallium; in addition, Dlx1/2 and Nkx2.1 are expressed in the striatum (Str) and pallidum, respectively [3]. The expression of homeobox genes provides evidence to determine the potentially homologous brain regions in vertebrates [4‒7].

Based on the expression of homeobox genes, fish and amphibians have relatively small pallium, while birds and reptiles have some expanded ones but are quite different from each other. First, the palliums of birds and reptiles are nuclear structures, while the dorsal pallium (neocortex) of mammals is a six-layered structure, with each layer composing a specific neuronal subtype. Layer-specific neurons are generated in a time-dependent inside-out gradient [8‒15]. Unlike mammals, neurogenesis follows a combination of outside-in and inside-out gradients in the avian dorsal pallium (Wulst, W), and an outside-in gradient in other two parts of pallium: mesopallium (M) and nidopallium (N), of them W and M/N molecularly and structurally belong to the mammalian dorsal neocortex and outside/ventral pallium, respectively [3, 16‒20]. It has been shown that a subset of radial glial progenitors that generates only upper-layer neurons and contributes to the development of callosal neurons (which are positive for Emx2+) in the developing mammalian brain is absent in the avian brain, and there are no homologous delayed radial glial progenitors in any regions of the chick pallium [13, 21]. Thus, from an evolutionary point of view, the mammalian neocortex is a novel structure [22, 23]. However, it remains unclear how this novel structure occurred.

Mice with mutations in homeobox genes exhibit developmental defects [24‒31]. Interestingly, some defects display characteristics that are similar to those of the nonmammalian telencephalon, including poor pallial lamination, a lack of the corpus callosum, and fewer cells expressing reelin, which can maintain the radial glial scaffold, thereby helping the neural progenitor cells to migrate to the destination [32‒34]. These data raise an interesting question regarding the degree to which the innovation of the mammalian neocortex is related to homeobox genes.

To address the above issue, we recently compared the coding sequences of genes vital for the development of the pallium (Emx1/2 and Pax6) and subpallium (Dlx2 and Nkx1/2) among 500 vertebrate species and found that these genes have no obvious variations in chromosomal duplication/loss, gene locus synteny or Darwinian selection. However, there is an additional fragment of approximately 20 amino acids in mammalian Emx1 and a poly-(Ala)_6–7 in Emx2, except those in the monotremes and marsupials in which the corpus callosum is lacking, as in Emx1/2-deficient mice [35]. These data are consistent with the fact that the subpallium is phylogenetically conserved compared to the mammalian neocortex. After transduction of mouse Emx2 lentiviruses into the ventricles of chick telencephalons at embryonic day (E) 3, the cells expressing reelin, vimentin, or GABA increased, and neurogenetic patterns of pallium changed toward mammalian inside-out one in some (but not all) pallial areas [35]. These data suggested that the pallium differences between birds and mammals could not be explained only by Emx2 variance.

Emx1 and Emx2 play a synergistic role in brain development, and defects are much more severe in Emx1/2 double mutants than Emx1 or Emx2 single mutant mice [24, 26, 27, 36, 37]. To further explore the roles of Emx1 and Emx2 in the organization and evolution of pallium in amniotes, lentiviruses containing mouse Emx1 and Emx2 (mixed with 1:1) were injected into the ventricle of the chick telencephalon at E3, and the embryos were allowed to develop to E12, E16, and posthatchling. In contrast, the injections of lentiviruses containing chick Emx1 or Emx2 (cEmx1/2) or no insert genes were as controls. Cells expressing reelin, vimentin, or GABA, neurogenesis patterns of calbindin (CB) and parvalbumin (PV) cells, and the changes in the anterior commissure (AC) immunoreactive for neurofilaments were examined and compared in the studied groups.

The eggs used in this experiment were purchased from Beijing Weishenghe Laboratory Animal Science and Technology Co., Ltd. All experiments in this study were conducted in accordance with the guidelines of Beijing Laboratory Animal Welfare and Ethical Review, and they were approved by the Animal Management Committee of the College of Life Sciences of Beijing Normal University.

Mouse and chick Emx1 or mEmx2 were synthesized according to the sequences in NCBI (Mouse: Emx1, NM_010131.2; Emx2, XM_004177159.1; Chick: Emx1, XM_001232150.7; Emx2, XM_025152057.2) and inserted into the pLVX-mCMV-ZsGreen vector (Shen Zhen BaiEnWei) to generate pLVX-m-Emx1 or Emx2-mCMV-ZsGreen1 lentiviruses (mEmx1 or mEmx2 Lv), pLVX-c-Emx1 or Emx2-CMV-ZsGreen1 lentiviruses (cEmx1 or cEmx2 Lv), respectively. The pLVX-mCMV-ZsGreen1 lentiviruses (Cv) lacking a targeted gene insert were used as a control. The viral titers were 2.0 × 10⁹ TU/ml.

Fertilized eggs were incubated in an incubator (99.5°F and relative humidity of 60–70%) for 72 h. After the eggs were sterilized using 70% alcohol, an 8-by-8-mm window was drilled at the top of the egg shell with a hand-held rotary tool. Under a stereoscopic microscope (Scoptic, China), the shell membrane was removed, and the vesicle of the telencephalon was observed over the eye and the beating heart. A glass electrode with a diameter of 15–20 μm and a pressure syringe were used to inject 3 µL of lentiviruses containing mouse Emx1/Emx2 (1.5 µL of mEmx1 Lv and 1.5 µL mEmx2 Lv), lentiviruses containing chick Emx1/Emx2 (1.5 µL of cEmx1 Lv and 1.5 µL cEmx2 Lv), lentiviruses containing no target genes or 3 µL of sterile saline into the telencephalic ventricle over 15–25 min. Before injection, the lentivirus liquid was diluted 20:1 with 0.1% fast green for an indicator (Sigma). The injection was considered successful if the green viral solution spread only in the vesicle of the telencephalon. After the injection, the eggs were placed blunt end up and returned to the incubator for further incubation. The embryos were allowed to develop to E12, E16, and to post-hatchling after 20 days of incubation. In general, approximately 75% of fertilized eggs were successfully hatched out following the above operation.

To date neuronal birthdays, a single dose of BrdU was injected into an area adjacent to a large vessel of the hatching eggs on consecutive embryonic (E) days (E4–E8) when most neurons in the pallium were produced. According to the embryo sizes, the BrdU (Sigma, 50 mg/mL) doses were 1 µL for E4, 1.3 µL for E5, 2 µL for E6, 3 µL for E7, and 6.8 µL for E8.

Total RNA was isolated from chicken telencephalons at E8 using TRIzol (Invitrogen) after injection of lentiviruses into the vesicles at E3. cDNA was generated using TIANScrip RT Kit reagents (TIANGEN, China) and amplified using the following gene-specific primers: Emx1 forward, 5′-GTA​TCC​GCA​CCG​CCT​TCT-3′; Emx1 reverse, 5′-CAC​TTT​CAC​CTG​GGT​CTC​G-3′; Emx2 forward, 5′-TTT​TGC​ACA​ACG​CTC​TGG​C-3′; Emx2 reverse, 5′-CTG​CTT​CCT​TTC​CGC​TCC​C-3′; β-actin forward, 5′-TTG​GCA​ATG​AGA​GGT​TCA​GGT-3′; and β-actin reverse, 5′-TAC​GGA​TGT​CCA​CAT​CAC​ACT-3′. The cycling conditions were as follows: 94°C for 30 s; and 40 cycles of 94°C for 5 s and 60°C for 30 s. The resulting fragment sequence was consistent with the target sequence. Each qRT-PCR assay was repeated three times. β-actin expression was used as a normalized control, and the normalized expression of the detected genes relative to β-actin was calculated by the 2-Ct method.

E12 and E16 brains were fixed in 4% paraformaldehyde without perfusion. P9 chicks were anesthetized with sodium pentobarbital and sacrificed by heart perfusion with frozen 0.1 m phosphate buffered saline (pH 7.4) followed by 4% paraformaldehyde. The perfused brains were stored in the same fixative at 4°C for 12 h and then soaked overnight in 30% sucrose. The brains were embedded in Jung compound (Leica) and cut into 10 μm thick slices with a freezing microtome (Leica CM 1850). The slices were placed on glass slides precoated with polylysine and stored at −20°C for Nissl staining or immunohistochemistry.

For DAB staining, the brain sections were first blocked with a 3% H₂O₂-methanol solution (diluted from 30% H₂O₂) for 25 min and then incubated with 5% donkey serum or rabbit serum in osmotic/blocking buffer for 30 min at room temperature. The sections were incubated overnight at 4°C with rabbit anti-MAP2 (Chemicon, 1:400), rabbit anti-GABA (Sigma, 1:1,000), mouse anti-reelin (Chemicon, 1:300), or mouse anti-neurofilament (Chemicon, 1:600). After the sections were washed, secondary antibodies conjugated to biotin, either goat anti-rabbit IgG (Zhongshanjinqiao Co., Ltd., 1:200) or horse anti-mouse IgG (Zhongshanjinqiao Co., Ltd., 1:200), in 0.5% Triton X-100/PBS were added at room temperature. The sections were then incubated with an avidin-biotin-peroxidase complex (Elite ABC kit, Vector, 1:150) for 2 h at room temperature and visualized with 3,3′-diaminobenzidine (DAB; Sigma, St. Louis, MO, USA) for 10 min.

For immunofluorescence staining, the sections were directly incubated with 5% donkey serum in osmotic/blocking buffer for 30 min at room temperature followed by incubation with the following primary antibodies overnight at 4°C in blocking buffer: mouse anti-vimentin (Chemicon, VM 3B4, 1:200), rat anti-BrdU antibody (AbD Serotec, 1:2,000), rabbit anti-CB (CB38; Swant, 1:500), or mouse anti-PV (Chemicon, MAB1572, 1:500). The sections were then incubated with donkey anti-mouse IgG (H + L) conjugated to Alexa Fluor 488 (Molecular Probes, 1:100), donkey anti-rat IgG (H + L) conjugated to Texas Red (Jackson, 1:400) or DyLight 488-conjugated sheep anti-rabbit IgG (Jackson ImmunoResearch, Cat. 313-486-003, 1:400) corresponding to the primary antibody.

For BrdU immunohistochemistry, the sections were pretreated with 2 N HCl for 3 h followed by incubation with Borax buffer (0.1 m, pH 8.4) for 30 min. In the control experiments, the primary antibody or the secondary antibody was omitted, but all other immunohistochemical procedures were the same as those described above. No staining was observed in these sections. According to the manufacturers, the above antibodies can be used in broad vertebrate species, including the chick, and the specificity of the primary antibodies was verified in previous reports and our preliminary experiment [35, 38, 39].

After 5-day-old chicks were labeled using footprints, they were placed in pairs in a pen (20 × 15 × 15 cm) 1 day before training for adaptation. The pens were illuminated by an overhead light (25 W), and the temperature was maintained at 25–30°C. The test procedure included pretraining, training, and tests (short-term and long-term memory tests). For the pretraining, the chicks were pretrained using a single presentation of a small white bead (2 mm in diameter) coated with pure water for 10 s. The above pretraining was repeated after 20 min. At 30 min after the pretraining, the chicks were subjected to a single presentation of a small red bead (3 mm in diameter) coated with pure water for 10 s. The number of times the bead was pecked was recorded as the background number (BN). After 30 min of pretraining, the chicks were given a red bead (same size as that used in pretraining) coated with methyl anthranilate (providing a bitter taste) for 10 s for the training. The number of times the chicks pecked the bitter bead was recorded. The chicks that pecked the bead and showed a disgusted reaction were considered to be trained. After 5 min (short-term memory test) or 120 min (long-term memory test) of training, chicks were given the same red beads (but dry) for 10 s, and the pecking times were recorded as peck (PN). The avoidance ratio (AR) was calculated using the following formula: AR = BN/(BN + PN). The AR was compared among the studied groups using the χ² test of independence.

To assess locomotor activities and curious or exploratory behaviors of the chicks, open-field tests were performed on three successive days (post-hatching days 5–7). The tests were performed in an open field (77 × 56 × 32 cm) formed with dark Formica flooring in a room illuminated by a 60-W light located 1.75 m above the center of the open field. A video camera fixed above the center of the open field was used to record the experimental sessions, and the recorded data were analyzed by Xeye Fcs (Beijing, China). Before the test, the chicks were coded, and the researchers were blinded to the codes until all the analyses were completed. Each chick at P5 was placed in the center of the open field and allowed to freely move for 15 min. The open field was cleaned with 75% ethanol after each test. A central rectangle corresponding to 38 cm × 28 cm in the open field was boxed from the image of the open field on the television screen. The following three measurements were examined on three successive days: the total distance of movement, the percentage of time spent in the central area and the number of entries into the central area.

Bright field images were acquired using a digital camera (Spot Enhance 2e; Diagnostic Instrument, Corp., USA) attached to a BH-2 microscope (Olympus, Japan). Immunofluorescent images were acquired using a ZEISS inverted fluorescence microscope (AxioCam MRm, Zeiss) under a ×20 or ×40 objective lens that was equipped with a monochromatic digital camera (AxioCam MRm, Zeiss). AxioVision Rel. 4.8 acquisition and processing software was used to acquire uniform digital images, and the images were converted to TIFF files and then analyzed using ImageJ and Adobe Photoshop.

The brain level and the delineation of regions were mainly referred to the chicken atlas [40]. Given that the present study was largely concerned with the changes among the W, M, N, and Str, the brain levels (from A 6.2-A 10.2 in the chick brain atlas) containing the above regions were selected. If the results had no significant differences among these brain levels, they were combined.

The data were compared using the SPSS 11.5 software package. The distributions of dependent variables were tested for normality, and homogeneity of variance was assessed for equality of error variance (Levene’s test). One-way ANOVA was used to compare differences between the experimental and control groups. As there were no significant differences among the groups receiving 3 µL of lentiviruses containing chick Emx1/Emx2 (1.5 µL of mEmx1 Lv and 1.5 µL mEmx2 Lv), lentiviruses with no targeted gene insert or 3 µL of sterile saline, the following results were only shown those from the group receiving 3 µL of lentiviruses containing mouse Emx1/Emx2 and those from the group receiving lentiviruses with no targeted gene insert for clarity. Data are presented as the mean ± SEM. Differences between the two sets of data were considered nonsignificant (p > 0.05) or significant at all levels (*p < 0.05, **p < 0.01, and ***p < 0.001).

In Emx2 mutants [24, 27] and Emx1/2 double mutants [26‒28], the distribution of reelin-expressing cells and neurogenetic patterns are largely altered, and the laminar structures of the pallium are disordered as shown by GABAergic interneurons or MAP2 cells, which specifically identify postmitotic cortical neurons. In addition, the neural connectivity between the cerebral hemispheres is malformed in Emx1 or Emx1/2 double mutants [24]. Therefore, we examined whether the AC is normally formed in the studied groups at posthatch day (P) 9.

Following the injection of 3 μL of lentiviruses containing mouse Emx1/Emx2 (1.5 µL of mEmx1 Lv and 1.5 µL mEmx2 Lv)or lentiviruses containing no target genes, the infected cells expressing GFP fluorescence were distributed uniformly in the W, M, N, and Str (Fig. 1a–d). The similar distribution of infected cells was also found after the injection of 3 μL of lentiviruses containing chick Emx1/Emx2 (1.5 µL of mEmx1 Lv and 1.5 µL mEmx2 Lv) (data not shown). qRT-PCR analysis showed a significant increase in the mRNA expression levels of mEmx1 and mEmx2 in the telencephalons collected at E8 after injections of mEmx1/2 Lv into the ventricle at E3 compared to those after injections of Cv (Fig. 1e, n = 5, mEmx1: p = 0.002; mEmx2: p = 0.008).

To examine the effects of mEmx1/2 Lv on the development of chick pallium (W, M, and N), the changes mentioned above were examined in the three brain levels, in which W, M, and N are presented, corresponding to A 10.2, A 8.2, and A 6.2 in the chick brain atlas, respectively [40]. Because the results did not significantly differ among the three brain levels, they were combined and not addressed. To study the neurogenetic patterns along the inside to outside gradient, the inside and outside regions of W, M, and N (400 × 600 m²) were selected, which were approximately 1/4 or 3/4 of the distance from the ventricle to the outermost side of the brain, respectively (Fig. 1d).

After injection of mEmx1/2 Lv into the ventricle, we examined the expression of reelin in W, M, and N at E16 when the development of the brain was complete [16]. Reelin-expressing cells increased significantly in both the inside and outside regions of W and M but only in the inside region of N (Fig. 2, n = 8, for each group; p < 0.01), whereas there was no significant change in the reelin-expressing cells in the outside region of n (Fig. 2d, h) compared to that in the group infected with control lentiviruses (n = 8, p = 0.569).

Given that vimentin guides cell migration [38, 41] and is expressed in radial glia processes, where Reelin maintains radial glial scaffolds [39, 41], we next examined the expression of vimentin in the W, M, and N. Vimentin expression was observed from E6 onwards but decreased after E14 and stopped after hatchling. Vimentin-labeled fibers initiated from the ventricle and then extended radially outward. However, the extending length of vimentin-labeled fibers varied in the W, M, and N. To quantitively assess how the vimentin-labeled fibers extended outward, vimentin-labeled fibers from the ventricle were traced continuously until they were interrupted, and the end of the fiber extending farthest was labeled with a dot (Fig. 3a–c). The sizes of the areas shaped by the ventricle and the labeled ends (S) to the length of the corresponding ventricular zone (L) (S/L) were then compared in the W, M, and N at embryonic day (E) 12 when a large number of cells were migrating toward their destinations. The results indicated that the S/L was higher in the W and M in the group infected with mEmx1/2 Lv than in the group infected with Cv (Fig. 3, n = 8, for each group; W: p = 0.007; M: p = 0.023). However, there were no significant differences in N (Fig. 3c, n = 8, for each group; p = 0.562).

GABAergic cells were significantly increased in the inside regions of W and M and outside region of W after injection of mEmx1/2 Lv into the ventricle at E3 (Fig. 4a, b, d; n = 8, for each group; p < 0.05). However, there were no significant changes in the inside region of N and the outside regions of M and N (Fig. 4c, e, f; n = 8, for each group; p > 0.05).

The numbers of MAP2-positive cells were significantly higher in the inside region of the inside W, M, and N in the group infected with mEmx1/2 Lv compared to the group infected with Cv (Fig. 5a–c; n = 8, for each group; W: p = 0.035; M: p = 0.008; N: p = 0.042). Moreover, there were no significant changes in the outside regions of the W, M, and N (Fig. 5e, f, g; n = 8, for each group; p > 0.05).

Previous studies have shown that cortical neurons, including GABAergic interneurons, such as CB and PV neurons, follow an inside-out neurogenetic pattern [42‒45]. Therefore, we investigated the pattern of neurogenesis in the W, M, and N for CB and PV neurons.

Following the injections of mEmx1/2 Lv or Cv into the ventricle of the telencephalon at E3 and BrdU injection on several consecutive embryonic days from E4 to E8, cells double-labeled for BrdU and CB or PV neurons were examined in the inside and outside areas of the W, M, and N at E16 (late stage of hatchling). Cells labeled for BrdU or double-labeled for BrdU and CB or PV neurons were observed in the groups injected with BrdU from E4 to E8 (Fig. 6; online suppl. Figure; for all online suppl. material, see https://doi.org/10.1159/000543601). The largest percentage of double-labeled cells for BrdU and CB or PV neurons to single-labeled cells for BrdU varied between the inside and outside areas examined and among the W, M, and N, suggesting different neurogenesis gradients in these areas. Therefore, we compared the largest percentages of double-labeled cells for BrdU and CB or PV neurons in the inside and outside regions of the W, M, or N. The results indicated that the largest percentages of double-labeled cells for BrdU and CB or PV neurons in the inside areas of the W or M appeared earlier than those in the outside areas after injection of mEmx1/2 Lv, suggesting an inside-out gradient of neurogenesis. In contrast, there was an outside-in neurogenesis gradient in the M following the injection of Cv (CB: Fig. 6a–j; PV: online suppl. Fig. S1A–J). Neurogenesis followed an outside-in pattern in the N in the two groups receiving the injection of mEmx1/2 Lv or Cv (CB: Fig. 6e–j; PV: online suppl. Fig. S1E–J).

Previous studies have shown that part of the AC of Emx1^−/− mice becomes disordered; however, the AC of Emx2^−/− mice does not change, but the latter part of the AC is deleted [3]. Therefore, we examined the sizes of AC immunostained for neurofilament across the midline of brain with a 1 mm length and compared the sizes between the two studied groups. The AC sizes were significantly larger in the group infected with mEmx1/2 Lv than in the group infected with Cv (Fig. 7a–c, n = 5 for each group).

Passive avoidance tasks were utilized to assess short- and long-term memory, while locomotor activities and curious or exploratory behaviors were examined using open-field tests. Following the injection of mEmx1/2 Lv into the ventricle of the telencephalon at E3, chicks at P5 showed a higher rate of passive avoidance at 120 min after training, but the rate had no significant changes at 5 min after training (Fig. 7d) compared to the group infected with Cv (n = 5 for each group; Fig. 7d). There were no significant differences in the total distance traveled by the animals and the percentage of time spent in the central rectangle between the two groups infected with mEmx1/2 Lv or Cv (Fig. 7e, f).

The present results indicated that injection of mEmx1/2 Lv into the ventricle of the chick telencephalon at E3 affected the development of the chick pallium, including increases in the expression of reelin, vimentin, GABA, and MAP2 in most areas examined in the W and M. Compared to previous results that were obtained by injecting only mEmx2 Lv [35], increases in the expression of reelin and GABA were observed not only in the inside part of M but also in the outside part of M. These results are largely consistent with previous reports to show that the distribution patterns of GABAergic interneurons, MAP2 and glutamate cells are disordered in Emx1/2 double mutants [26‒28]. We also attempted to study glutamatergic cells by using antibodies available including anti-Glu. However, ideal results were not obtained due to the lack of strong, specific immunohistochemical staining.

Our results indicated that the AC sizes were significantly larger after infection with mEmx1/2 Lv, but there were no significant differences in AC size after infection with only mEmx2 Lv [35]. Thus, the present results supported the view that Emx1 and Emx2 have a synergistic role in the development of cortical organization and neural connectivity between the two cerebral hemispheres [24‒27, 36, 37].

As mentioned above, the mammalian neocortex is a novel structure, and the corpus callosum is only present in Eutherian species. However, the mechanisms underlying the evolution of these structures are not completely known. As mentioned above, our recent study has shown that there are no obvious variations in chromosomal duplication/loss, gene locus synteny or Darwinian selection for the coding regions of genes for Emx1/2, Pax6, Dlx2, and Nkx1/2, but there is an additional fragment of approximately 20 amino acids in Emx1 and a poly-(Ala)_6–7 in Emx2 in all of the examined mammals [35]. Emx2 contains only two homopolymeric alanines in the examined non-Eutherian species, including monotreme (duckbill platypus) and marsupials (gray short-tailed opossum and Tasmanian devil), in which the corpus callosum is lacking as in nonmammal species.

Given that Emx1 plays some indispensable roles in the development of the pallium, including the normal establishment of neural connectivity between the two cerebral hemispheres, especially the corpus callosum in Eutherian species, and Emx1 and Emx2 have synergistic actions in the development of the brain [25, 27], the lack of a corpus callosum in non-Eutherian species is regarded to be related to the additional fragment of approximately 20 amino acids in Emx1 and the 6–7 homopolymeric alanines in Emx2 [35]. Although the corpus callosum is substantially lacking in the avian pallium and there are large differences in pallium organization between birds and mammals, the present result is consistent with previous reports on Emx1 or Emx2 mutant mice indicating that there are significant effects on AC after changes in both mEmx1 and mEmx2 expression but no obvious effects after only changes in mEmx2 expression in chicks [25, 27, 35]. The present study confirmed the role of Emx1 in the normal establishment of neural connectivity between the two cerebral hemispheres [25, 27]. Therefore, these findings suggested that Emx1 and Emx2 are involved not only in the development of the pallium but also in the evolution of neural connectivity between the two cerebral hemispheres. However, this inference does not exclude the possibility that other genes are also involved.

At least a part of Cajal-Retzius and reelin cells are reported to originate from the inside ganglionic eminence (Lhx6 positive) [46], whereas GABAergic interneurons originate from the outside ganglionic eminence and migrate tangentially into the cortex [46‒48]. Our results indicated that both reelin and GABA cells increased after the injection of mEmx1/2 Lv into the ventricle of the chick telencephalon at E3, suggesting that mEmx1/2 may affect the final positions in the pallium of these cells. However, the mechanism remains unclear, and it needs to be further studied.

Reelin expression is detected in Cajal-Retzius cells, but not in the GABAergic neurons, in the marginal zone (layer I) of mouse cerebral cortex from E10 to 12.5, and later appears in cells in the cortical plate from E14 to P5 (not restricted only in Cajal-Retzius cells), and largely decreases from P5 onwards [49]. This complex pattern of cellular and regional expression of reelin suggests that reelin may play multiple roles in brain development and adult brain function. Our present results indicated that cells expressing reelin increased significantly at E16 in both the inside and outside regions of W and M, but only in the inside region of N, after the injection of mEmx1/2 Lv into the ventricle of the chick telencephalon. However, reelin expression is largely reduced after post-hatching. Thus, reelin expression is corresponded to the developmental stages of pallium, supporting its roles involved in the development of chick pallium.

According to a previous report [39], after E7 pallium slices of quail are co-cultured with COS7 cells expressing reelin, straight projection of GFP-labeled radial fibers extending from radial glial cells are directed toward the source of reelin, whereas co-cultured without reelin-positive cells, radial glial cells extend fibers in multiple orientations. This report suggests that reelin controls directed growth of radial fibers in the quail pallium as in mammals [28]. The corresponding stages were also studied in the present report, and we thus infer that the related role of reelin, as reported in the quail pallium [28], also plays in the development of chick pallium. However, the roles of reelin in the development of avian pallium requires to be further studied.

Studies on mammalian species indicate that reelin acts as a chemoattractant or a positional signal to guide the radial migration of cortical cells toward the marginal zone [50, 51], and the lack of normal reelin protein, as in reeler mice, causes abnormal layering of the cortical plate in which newborn neurons are unable to bypass preexisting populations of older neurons, resulting in an inverted outside-in pattern. In addition, reelin is largely affected by Emx2. Reelin cells are predominantly absent in neocortex and paleocortex regions in Emx2 single mutants, but they are completely absent in the superficial portion of the entire cerebral cortex in Emx1/2 double mutants [28]. A study on the developmental origin of avian cortical neurons has demonstrated that corticogenesis instructions produced by reelin-expressing CR cells and subplate cells are absent in birds [52]. Thus, the variances in neurogenesis gradients in the pallium of birds and mammals may be related to the difference in the coding sequences of Emx2 as described above.

Reelin attracts radial glial fibers (vimentin⁺) to extend vertically to their destination in the mammalian cortex [37, 53‒55]. The present study indicated that vimentin expression was largely consistent with the development of chick pallium (observed from E6 onwards, decreased after E14 and stopped after hatchling), and vimentin-labeled fibers initiated from the ventricle, and extended radially outward. Thus, the extending direction of Vimentin-labeled fibers are orientated toward the destination of migrating cells, and these are consistent with previous reports on birds or the reptile, indicating that some cells were observed to migrate along vimentin fibers [38, 41, 56]. These results suggest that vimentin fibers are most probably involved in the guide of cell migration toward their target destinations. However, it remains unknown whether this guide of cell migration is related to the neurogenesis pattern (outside-in or inside-out).

In the present study, reelin-expressing cells were significantly increased in both the inside and outside regions of the W and W but not in the outside region of the N (Fig. 2), and Vimentin-labeled fibers extended radially much further from the ventricle zone in the W and M but not in the N in the chicks infected with Emx1/2 Lv. Correspondingly, GABA-expressing cells were present in most areas examined in the W and M after injection of mEmx1/2 Lv into the ventricle of the telencephalon. It has been shown that GABA cells are affected by reelin, as their numbers decrease and they cannot normally migrate to the cerebral cortex, resulting in abnormal arrangement in the cerebral cortex as shown in reeler, Emx2^−/− or Emx1/2 double mutant mice [26]. Thus, the present study supports the previous report to show that the production of GABA is affected by Emx1/2.

The distribution of MAP2 in cortical layers is normal in Emx1^−/− homozygotes, but the MAP2-negative superficial layer is thin in Emx2^−/− homozygotes, and absent in Emx1/2 double homozygotes. The present study indicated that MAP2 neurons were significantly increased in the inside region of the W and M after infection with mEmx1/2 Lv, but there were no significant changes in the W and M after infection with mEmx2 Lv (Fig. 5), strengthening the synergistic action of Emx1 and Emx2.

Our results indicated that neurogenesis pattern of cells for CB or PV changed in the W and M, but not in N in the group infected with mEmx1/2 Lv. Previous reports indicated that neurogenesis in W partly follows the inside-out pattern, but it follows the outside-in pattern in M and N [16, 17, 20], whereas mammalian interneurons follow an inside-out pattern of neurogenesis [44, 45]. The above results are consistent with those to indicate that reelin, vimentin and GABA cells changed in the W and in the inside part of M but not in the N (the ventral pallium lacking Emx1 expression) [3].

Although no significant differences in locomotor activity were observed in the open-field tests, the passive AR was significantly increased in the group infected with mEmx1/2 Lv. It has been shown that the medial intermediate medial mesopallium (IMM) in the M is a site involved in memory formation in chicks, the learning and memory ability of chicks is closely related to the neuronal function of IMM [57, 58]. Thus, it is possible for the increase in GABA neurons in the medial part of the M to be related to the changes in the passive AR, during long-term memory storage, mEmx1/2 may indirectly cause changes in other specific neuronal subtypes or circuits by regulating the expression of GABA neurons.

Importantly, some changes appeared in the palliums of chicks after infection with mEmx1/2 Lv, and these changed features varied from those of the mammalian cortex; however, no significant changes occurred in some regions of the chick pallium, such as the N. These differences may be explained by several reasons. First, by injecting mEmx1/2 Lv into the ventricle of the chick telencephalon at E3, a chimeric brain was obtained as only some cells were infected with lentiviruses expressing target genes, which may lead to the above results that brain development was only partly affected. To date, transgenic birds are not available because it is difficult to obtain a single pronucleus from a large yolky oocyte [59, 60]. Nevertheless, chicken embryos are widely used in developmental biology and neuroscience, owing to the ready availability of fertile eggs and early publication of chick genome sequences [61, 62]. Second, it should be noted that the development and evolution of the cerebral cortex is complex and has evolved over millions of years in vertebrates. Thus, the development should not be ascribed to only Emx1/2, and the development and evolution of the cerebral cortex may be involved in a large network of genes. The mechanisms under which the cerebral cortex develops and evolves among vertebrates should be further studied.

Huiru Yan performed some studies in the earlier stage of this study.

This study protocol was reviewed and approved by Animal Welfare and Ethics Committee, College of Life Sciences, Beijing Normal University, Approval No. 2021-SW-013.

The authors declare that no competing interests exist.

This study was supported by grants from the National Natural Science Foundation of China to S.J. Zeng (Nos. 31172082, 31970414 and 32470493).

Conceived of and designed the experiments: S. Zeng and X. Zhang; performed the experiments: R. Zhao, Y. Gao, P. Liu, S. Lin, and S. Lu; analyzed the data: C. Xi, J. Liu and J. Bing; drafted the manuscript: S. Zeng, R. Zhao, and C. Xi.

Rui Zhao, Yuanyuan Gao, Chao Xi, and Ping Liu contributed equally to this work.

All data that support the findings of this study are included in this article and its online supplementary material. Further inquiries can be addressed to the corresponding author.