Experimental procedures were performed in accordance with the guidelines of the Animal Care and Use Committees at the Shenzhen Institute of Advanced Technology (SIAT), Chinese Academy of Sciences (CAS), China (permit number SIAT-IACUC-20210607-NS-NBJZX-ROBERT NAUMANN-A1706-01). Wild type pregnant Sprague Dawley (SD) rats were supplied by Beijing Vital River Laboratory Animal Technology Co., Ltd. Both male and female rats aged from E13.5 to postnatal day 20 (P20) were used in this study. Food and water were given ad libitum. The day females were sperm positive was designated as E0.5. Pups were typically born on E21.5-E22.5, and the birth day was designated as P0.

5-ethynyl-2′-deoxyuridine (EdU, ApexBIO, Houston, United States, Cat# B8337) was diluted in 0.1 M phosphate buffered saline solution (PBS). Female timed-pregnant rats (N = 18, 3 per group, bodyweight 280–320 g) received a single 25 mg/kg EdU injection given between 17:00 and 18:00 on days corresponding to E12.5 to E17.5. At birth, all newborn rats were perfused.

Pregnant rats or rat pups (P0, P2, P4, P8 and P20) were briefly anesthetized with isoflurane and subsequently given an intraperitoneal injection of 60 mg/kg pentobarbital. Rat pups were perfused with 0.1 M DEPC (diethyl pyrocarbonate)-PBS and 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer (PB), followed by brain dissection and fixation with 4% PFA overnight. Embryonic brains were dissected out in ice cold DEPC-PBS, followed by fixation in 4% PFA overnight. Before sectioning, rat brains were cryoprotected using a 30% sucrose solution in DEPC-PBS for over 24 h. We used OCT compound for embedding brains and prepared 50 μm thick coronal sections.

In situ hybridization was performed using a protocol based on Ferran et al. (2015). DIG-labeled Riboprobes were used for hybridization on 50 μm free floating cryosections. Hybridization was performed overnight at 65°C without proteinase K treatment. Sections were washed at 65°C two times in 2xSSC/50% Formamide/0.1% N-lauroylsarcosine for 20 min, twice in 2xSSC/0.1% N-lauroylsarcosine at 37°C for 20 min and twice in 0.2xSSC/0.1% N-lauroylsarcosine at 37°C for 20 min. Sections were blocked in MABT/10% goat serum/1% Blocking reagent (Roche Diagnostics, Mannheim, Germany, Cat# 11096176001), incubated overnight with sheep anti-DIG-AP (1:1000, Roche Diagnostics, Mannheim, Germany, Cat# 11093274910). After washing, staining was performed using NBT/BCIP in NTMT until satisfactory intensity was reached. Staining reaction was stopped with 10 mM EDTA. Sections were washed, dehydrated and mounted with Eukitt^® Quick-hardening mounting medium (Sigma Aldrich, St. Louis, United States, Cat# 03989).

Fluorescent in situ hybridization was performed using DIG-labeled probes. After hybridization and washing, sections were first incubated with sheep anti-DIG-POD (1:1000, Roche Diagnostics, Mannheim, Germany, Cat# 11207733910) and tyramide signal amplification (TSA) was performed using biotin-tyramide (ApexBIO, Houston, United States, Cat#A8011). For TSA, sections were equilibrated 5 min in TSA buffer (10 mM imidazole in PBS), then incubated in the TSA staining solution (2 μg/ml biotin-tyramid/0.001%H₂O₂/TSA buffer) in the dark for 30 min without shaking. After washing, sections, subsequently, were incubated with Streptavidin-Cy2 (Jackson ImmunoResearch) to detect the DIG labeled probe. Sections were washed, then mounted with Fluromount medium (Sigma Aldrich, St. Louis, United States, Cat# F4680).

Rat cDNA was synthetized from total brain RNA using EasyScript First-Strand cDNA Synthesis SuperMix (Transgen Biotech, Beijing, China, Cat# AE301-02). The Nurr1 probe was amplified with the following primer pairs: Fw-tgttcaggcgcagtatgggtcc. Re-tcacaagtgcgaacaccgtaatgc, for a total length of 834 bp, and then cloned in a pEASY-Blunt Zero backbone (Transgen Biotech, Beijing, China, Cat# CB501-01) and verified by sequencing. Antisense digoxigenin-labeled riboprobes were synthetized according to the protocol recommended by the manufacturer (Roche Diagnostics, Mannheim, Germany).

Detection of EdU was performed after in situ hybridization with a modified protocol based on Salic and Mitchison, 2008 and Hong et al., 2009. Free-floating sections were washed 3 × 5 min in PBS, 1 × 20 min 0.5% Triton X-100/PBS, 3 × 5 min in PBS. Subsequently, sections were incubated for 30 min in the EdU click chemistry reaction buffer in the dark at room temperature. The reaction buffer (working solution) was prepared freshly by combining in the following order: (1) 0.1 M PBS (2) 5 mM CuSO4, (3) 1 mM THPTA (tris-hydroxypropyltriazolylmethylamine), (4) 30 mM Na-Asc (Ascorbic Acid Sodium Salt), and (5) 25 μM Cy3 azide (ApexBIO, Cat# A8127). When preparing the reaction buffer, mix well after adding each component, the solution should be transparent after adding Na-Asc. After the click reaction, sections were washed 3 × 5 min in PBS. A detailed ISH and EdU labeling protocol can be found online https://wangnaumannlab.cn/wj.

For quantification of cells expressing both Nurr1 and EdU, we use P0 sections where all Nurr1 labeled sub-regions can be clearly identified and are comparable across animals. For each individual region, we counted double positive cells from three animals in three equally distanced sections at the coordinates indicated below. For the ventral DEn (vDEn), dorsal DEn (dDEn), vCL, and dCL, we use anterior sections at bregma 0.8 to 0.6 mm. For the dLn, we use intermediate sections at bregma −0.4 to −0.6 mm. For the superficial layer neurons (sLn) and posterior DEn (pDEn), we use posterior sections at bregma −1.6 to −1.8 mm (coordinates based on Khazipov et al., 2015). We focus on injection dates with significant EdU expression for each subregion: E12.5 to E15.5 for the vDEn, dDEn and pDEn; E13.5 to E15.5 for the vCL and dCL; E13.5 to E16.5 for the dLn and E15.5 to E17.5 for the sLn.

All chromogenic and fluorescent stainings were imaged at 10 × or 20 × magnification on a slide scanner (BX61VS, Olympus, Japan) or at 20 × magnification on an ApoTome microscope (Axio Imager 2, ZEISS, Germany). The fluorescent images were acquired in monochrome, and color maps were applied to the images post acquisition. Post hoc linear brightness and contrast adjustment were applied uniformly to the image under analysis.

We study the birth dating of the claustrum, endopiriform nucleus, and related cortical regions that are labeled by Nurr1 gene expression by injecting EdU at different time points in timed-pregnant rats. We define E0.5 as the first day that females were sperm positive. We investigate the distribution of EdU labeling and Nurr1 expression in rat pups at birth P0 (Figure 1A). Since EdU toxicity may adversely affect cellular health (Andersen et al., 2013), we sought to identify the lowest dose of EdU that allows reliable birth dating while maximizing pup survival. We find that injections of 10 mg/kg or 25 mg/kg EdU show little effects on litter size, while higher doses correlate with reduced litter size (Figure 1B). The survival time after EdU injection was not correlated with litter size (Figure 1B). Thus, we subsequently use injections of 25 mg/kg for all further experiments. Figures 1C,D shows the lateral and dorsal regions of cortex of P0 rats after consecutively timed EdU injections. In the dorsal cortex (Figure 1C), we observe EdU signal in cortical layer 6 mainly in animals injected on E14.5, in layer 5 mainly in animals injected on E15.5 and in upper layers in animals injected on E16.5 and E17.5. In contrast, in the lateral cortex (Figure 1D) EdU signals appeared earlier and relatively few newborn neurons were detected in animals injected later than E16.5. To identify claustrum neurons we use in situ hybridization for Nurr1, a relatively selective and early expressed marker (Puelles, 2014). We study seven brain regions labeled by Nurr1 illustrated in three coronal sections of the rat brain at P0 (Figure 1E). In the anterior section, Nurr1 labels four sub-regions, including dCL and vCL. Since the DEn shows dark and light staining parts, we divide it into dDEn, and vDEn as suggested previously (Bayer and Altman, 1991; Arimatsu, 1994). In the intermediate section, we identify Nurr1 expression in cortical deep layers dLn. In the posterior section, Nurr1 is expressed in cortical superficial layers sLn, and the elongated pDEn. Another region containing Nurr1 positive neurons, the Sp has been studied in great detail (reviewed in Hoerder-Suabedissen and Molnár, 2015; Molnár et al., 2020), so we do not investigate it here.

During prenatal development, we observe the earliest Nurr1 expression in the prospective claustrum and dorsal endopiriform regions (CL/DEn) on E13.5 (Figure 2A). In intermediate sections, Nurr1 first appears as a small elongated region. Its ventral part is located closer - its dorsal part more distant to the brain surface (Figure 2Aa). In more posterior sections, we observe little Nurr1 staining in the lateral forebrain yet a large group of Nurr1 positive cells is present at more ventral locations marked with a star symbol. We observe a direct connection of this ventral posterior Nurr1 positive region and the prospective CL/DEn region marked by an arrow (Figure 2A). E15.5 (Figure 2B) sees a major shift in the intensity and distribution Nurr1 expression. Nurr1 now labels the CL/DEn in more anterior and more posterior sections than in earlier stages, also weakly labeled in the lateral part of the cortex. Sp neurons are also weakly labeled. Anteriorly, CL/DEn has reversed its orientation. Its dorsal part is now closer to the brain surface while the ventral part is relatively distant from the brain surface (Figure 2Bb1). In the intermediate sections the CL/DEn starts to divide into two parts with different staining intensity (Figure 2Bb2). The elongated shape of the pDEn becomes apparent in more posterior sections (Figure 2Bb3). From E17.5 (Figure 2C) until birth the Nurr1 expression pattern in the prospective CL/DEn shows less overall changes, but becomes more differentiated. Anteriorly, different subregions of CL/DEn are not yet apparent on E17.5 (Figure 2Cc1) but at more intermediate levels more subregions become apparent, we tentatively label them as vCL, dDEn, and vDEn (Figure 2Cc2). In posterior sections, the pDEn appears continuously with the Sp neurons (Figure 2Cc3), while at intermediate sections there is a small gap between CL and Sp neurons and in anterior sections, a wider gap (Figures 2Cc1,c2). In addition, Nurr1 starts to weakly label the sLn (Figure 2Cc3). On E19.5 (Figure 2D), Nurr1 staining becomes more differentiated, with the vCL and dCL subdivisions visible (Figure 2Dd2). The gap between the brain surface and the CL, DEn, and pDEn keeps widening. In intermediate sections, the dLn shows first staining, and in posterior sections the sLn shows stronger Nurr1 expression (Figures 2Dd2,d3). On E20.5 (Figure 2E), all sub-regions that we defined based on Nurr1 expression in P0 brains are readily apparent (Figures 2Ee1–e3). In summary (Figure 2F), on E13.5 Nurr1 first shows weak expression in a relatively ventral position. A major change occurs from E13.5 to E15.5, when Nurr1 shows stronger expression in an elongated stripe from anterior to posterior and subdivisions of different staining intensity become visible. In the sLn, Nurr1 shows weak expressions on E17.5 and increased expression on E19.5. On E20.5, Nurr1 strongly labels all claustrum sub-regions.

During early postnatal development Nurr1 expression in the claustrum and adjacent nuclei remains largely unchanged from its pattern at birth (Figure 1E). However, Nurr1 expression in the posterior cortex changes dramatically during early postnatal development (Figure 3A). Nurr1 expression in the sLn is greatly reduced from P2 to P20 (Figure 3B). On P2, Nurr1 has the strongest expression in superficial layers, while on P20, there is almost no Nurr1 expression in that region (Figure 3C). At birth, Nurr1 expression in dLn neurons shows some distance from the Sp, while on P20, Nurr1 expression in dLn is closer to the Sp (Figure 3C). We also note the approximately 3-fold expansion in cortical thickness taking place from P0 to P20 (see scale bars in Figure 3C). In summary, Nurr1 expression in the sLn differs from other Nurr1 positive populations we describe, in that its expression level is greatly reduced during early postnatal development (Figure 3D). Further, dLn neurons differ from other Nurr1 positive neurons in changing their apparent laminar position during early postnatal development. Around birth, they are located mostly in the upper half of the cortex, intermingled with sLn cells, while on P20 they are located in the deep layers of lateral cortical regions. We did not observe major changes in Nurr1 expression patterns after P20.

For each EdU injection date from E12.5 to E17.5, we selected 6 sections at comparable distances from anterior to posterior to show the EdU expression patterns combined with Nurr1 in situ hybridization and DAPI staining (Figures 4–9). High resolution images can be accessed here: https://figshare.com/s/9e8fc7cc66be7d9e968b. We briefly describe EdU labeling in selected regions for each injection day, while a more detailed description of EdU and Nurr1 co-labeling follows afterward. EdU injections on E12.5 (Figure 4) reveal mainly signals in the piriform cortex (Pir), intermediate endopiriform nucleus (IEn), anterior cortical amygdaloid area (ACo), basomedial amygdaloid nucleus, anterior part (BMA), medial amygdaloid nucleus (Me), medial amygdaloid nucleus, posteroventral part (MePV). Apparently, there are also some labeled cells in the incipient Sp at E12.5. Following EdU injections on E13.5 (Figure 5) we find additional EdU signals in the claustrum (CL), dorsal endopiriform nucleus (DEn), insular cortex (Ins), Sp, basomedial amygdaloid nucleus (BM), medial amygdaloid nucleus, posterodorsal part (MePD), basomedial amygdaloid nucleus, posterior part (BMP), amygdalohippocampal area, posterolateral (AHiPL), and posteromedial cortical amygdaloid area (PMCo). Following EdU injections on E14.5 (Figure 6) we find reduced EdU signal in the IEn, ACo, Me, and MePV. In addition EdU signals are visible in deep layers of secondary somatosensory cortex (S2), basolateral amygdaloid nucleus, anterior part (BLA) and the basolateral amygdaloid nucleus, posterior part (BLP). EdU injections on E15.5 (Figure 7) show reduced EdU in the DEn, deep layers of Ins, medial and posterior Pir, and the lateral part of Sp. In addition, EdU signals are visible in the superficial layers of Ins, intermediate layers of S2, secondary auditory cortex, dorsal area (AuD), and primary visual cortex (V1). Following EdU injections on E16.5 (Figure 8), EdU signals are reduced in the CL, superficial layers of Ins, intermediate layers of S2, and the dorsal part of Sp. In addition EdU signals are visible in superficial layers of AuD, V1. EdU injections on E17.5 (Figure 9) show reduced EdU signals in superficial layers of AuD but still show signals in the superficial layers of V1.

To study the birth dates of Nurr1 positive neurons, we first describe EdU signals at the level of subregions, focusing first on dCL, vCL, dDEn, vDEn (Figure 10). On E12.5 (Figure 10A), we find few EdU signals in the vDEn, dDEn, vCL, and dCL. On E13.5 (Figure 10B), we find greatly increased EdU signals in the dDEn and vDEn, with some signals in the vCL and dCL. We find that the ventral part of the vCL shows more EdU signals than its dorsal part. On E14.5 (Figure 10C), vCL shows more EdU signal, and more EdU is visible in the dCL. In addition, there are more EdU signals in the dorsal part of the dDEn than in its ventral part. On E15.5 (Figure 10D), vCL is still densely labeled with EdU signals, but with more EdU signals in the dorsal part than its ventral part. The dCL still has some EdU signal, but there is less EdU in the vDEn and dDEn. A small region located medial of the DEn known as the reservoir (Bayer and Altman, 1991), paraclaustral reservoir (Maciejewska et al., 1999), or bed nucleus of the external capsule/BEC (Puelles, 2014; Puelles et al., 2016b) shows dense EdU labeling in its ventral part but does not express Nurr1. On E16.5 (Figure 10E) and E17.5 (Figure 10F), few EdU signals are visible in the claustrum sub-regions, but the paraclaustral reservoir/BEC is densely labeled with EdU in its dorsal part (Figures 10E,F). In summary, EdU signals first appear in the vDEn and dDEn on E12.5, first in the vCL and dCL on E13.5, first in the paraclaustral reservoir/BEC on E15.5 (Figure 10G). In Figure 10H, we summarize the ventral to dorsal gradients in birth dating. vDEn neurons are mainly born on E13.5 and dDEn neurons are mainly born on E13.5 to E14.5, then most vCL and dCL neurons are sequentially born on E14.5 to E15.5, and paraclaustral reservoir neurons are born on E15.5 to E17.5.

For the remaining Nurr1 positive regions (dLn, sLn, pDEn), we study more posterior sections (Figure 11). On E14.5, we notice EdU signals mainly in the ventral parts of dLn (Figures 11Aa1,a2). On E15.5, we find EdU signals mainly in the dorsal superficial layer parts of dLn (Figures 11Aa3,a4). On E16.5, few EdU signals are in the dLn (Figures 11Aa5,a6). On E15.5, EdU signals are mainly in the ventral parts of sLn (Figures 11Bb1,b2). On E16.5 and E17.5, we observe EdU signals mainly in the dorsal parts of sLn (Figures 11Bb3–b6). On E12.5, there is little EdU signal in the pDEn. However, on E13.5 and E14.5, the pDEn is densely labeled with EdU (Figure 11C). In summary, EdU signals appear in the dLn mainly on E14.5 and E15.5, in the sLn mainly on E15.5 to E17.5, in the pDEn mainly on E13.5 and E14.5 (Figures 10G, 11D). However, we note that both dLn and sLn are intermingled with other Nurr1 negative cortical neurons, therefore, we proceed to directly quantify Nurr1 overlap with EdU.

We quantified the fraction of Nurr1 + neurons at P0 that were born on a particular embryonic day (Figure 12), by determining how many of the Nurr1 + neurons (green) co-localized (Nurr1 + EdU +) with EdU + neurons (red). In the vDEn, we find the largest fraction of Nurr1 + EdU + cells on E13.5 (57% ± 6.9). In the dDEn, most Nurr1 + EdU + cells are born on E13.5 and E14.5 (42% ± 6.3, 41% ± 4.5). Similarly, in the pDEn, most Nurr1 + EdU + cells appear on E13.5 and E14.5 (43% ± 5.8, 44% ± 4.4). In the vCL, most Nurr1 + EdU + cells are born on E14.5 and E15.5 (39% ± 2.5, 36% ± 2.5). Likewise, in the dCL, we find the largest fraction of Nurr1 + EdU + cells on E14.5 and E15.5 (36% ± 2.1, 35% ± 2.2). Also, in the dLn, we find most Nurr1 + EdU + cells on E14.5 and E15.5 (33% ± 2.2, 37% ± 2.9). In the sLn, most Nurr1 + EdU + cells show on E16.5 and E17.5 (44% ± 6.2, 34% ± 2.7) (Figure 12B). In summary, we find that all regions show double positive cells on more than one day. Most vDEn neurons are born by E13.5, followed by dDEn and pDEn on E13.5 to E14.5, vCL, dCL, and dLn on E14.5 to E15.5, and sLn on E15.5 to E17.5. Based on the birth dates of the sub-regions above, we find dDEn is most similar to pDEn, while vCL, dCL, and dLn are also similar. In contrast, vDEn and sLn show distinct neurogenetic patterns.

To investigate anterior to posterior differences in birth dates of CL and Den, we show four comparable sections with injections between E13.5 and E15.5 (Figure 13). We focus on CL and DEn, since further subdivisions are not easy to define consistently at all anterior posterior levels. On E13.5, we find less EdU signals in vCL anteriorly, but more EdU signals in DEn posteriorly. On E14.5, we find that vCL is densely labeled with EdU except in the most anterior section. We also observe decreased EdU signals in the ventral part of DEn. On E15.5, we find decreased EdU signals in the ventral part of vCL, and also decreased EdU signals in the dorsal part of DEn (Figure 13A). In summary, EdU signals appear in the posterior DEn mainly on E13.5 and E14.5, but in the anterior DEn mainly on E13.5, E14.5 and E15.5. EdU labels the posterior CL mainly on E14.5, but the anterior CL mainly on E14.5 and E15.5 (Figure 13B).

In the claustrum, Nurr1 expression has been reported in the prospective primordium of claustrum in the superficial part of lateral pallium as early as E12.5 in mice (Puelles, 2014). Rat embryonic brain development is thought to attain a comparable state about one to two days later than in mice (Schneider and Norton, 1979; Clancy et al., 2001). Consequently, we find the first Nurr1 staining in the lateral forebrain on E13.5. However, we note that Nurr1 staining on E13.5 in the rat shows some distance to the surface of pallium, with its ventral part relatively closer to the pallium surface than its dorsal parts (Figure 2Aa). Interestingly, while the prospective claustrum neurons remain relatively weakly stained at this time point, we find that there is a more strongly stained posterior-ventral Nurr1 positive region. By following a series of coronal sections (Figure 2A), or on sections cut parallel to the surface (Figure 14A), we find that both Nurr1 positive regions appear to be continuous.

While Puelles (2014) proposed that Nurr1 positive claustrum neurons are generated in the lateral pallium, a recent study by Moreau et al. (2021) finds that in the mouse on E12.5 Nurr1 positive neurons are located in the ventral pallium. One possibility is that these cells are unrelated to claustrum neurons and reduce their Nurr1 expression over time. However, if they are related to claustrum neurons, their appearance in different sectors of the pallium remains unexplained (Moreau et al., 2021). Cell lineage analysis supports the former view, suggesting that claustrum and endopiriform neurons are indeed generated in the lateral pallium (Garcia-Moreno et al., 2018; Rueda-Alaña et al., 2018; Bruguier et al., 2020). By E15.5, the majority of Nurr1 staining has congregated in a lateral stripe in the prospective CL/DEn region (Figures 2B, 14B). Little staining remains visible in the previously strongly labeled posterior-ventral parts of the pallium, indicating either loss of Nurr1 expression, or depletion of Nurr1 positive neurons in that region (Figures 2B, 14B). Consistent with the results of Puelles (2014), we find that over the next days of embryonic development the formerly uniform elongated Nurr1 positive region differentiates into multiple subdivisions (Figures 2B–D). Among those, the earliest visible subdivision is based on the light and dark staining of Nurr1 in the DEn, dividing it into ventral and dorsal parts (vDEn and dDEn, Figure 2Bb2). This segmentation of the DEn can also be observed in adult animals by using additional markers including Lxn, Ntng2, and Oprk1, which are enriched in dDEn, and Cpne4, Pcp4, and Npsr1, which are enriched in vDEn (Arimatsu, 1994; Naumann et al., in preparation). The molecular characterization of cell types in the endopiriform nucleus remains largely unexplored (Smith et al., 2019). However, identifying mouse lines with specific expression in subsets of DEn neurons (Riedemann et al., 2019) will enable elucidating their development, connectivity, and function.

In more dorsal parts of cortex, Nurr1 positive sLn neurons first appear faintly on E17.5 (Figure 2Cc3), while dLn neurons first become visible somewhat later, on E19.5 (Figure 2Dd2), comparable to descriptions of dLn in the mouse (Puelles, 2014; Ratié et al., 2020). By E20.5, both sLn and dLn neurons are found in a broad region throughout the posterior dorsal cortex (Figures 2E, 14C). Interestingly, sLn and dLn neurons show a sharply divergent postnatal development (Figure 3). In the sLn region, Nurr1 expression is strongly reduced over the first 10 days of postnatal development and essentially absent by P20. In contrast, dLn neurons become more strongly stained and appear to settle in deeper layers. This observation may be explained by a coordinated pattern of up - and down-regulation of Nurr1 expression involving multiple neuronal populations at different distances from the subplate during early postnatal development. An alternative explanation for such coordinated changes in Nurr1 expression was suggested by Li et al. (2011). Using antibodies against Nurr1, they observed similar changes in staining patterns in the cortex and embryonic spinal cord, interpreting the latter as evidence for neuronal migration. Interestingly, Nurr1 may have a role in neuronal migration during the development of midbrain neurons (Wallén et al., 1999), however, no evidence for such a function is available for Nurr1 in forebrain neurons. In addition, we note that, while the generation of the laminar pattern of cortical neurons is essentially complete before birth (Figures 1C,D), cortical thickness around dLn neurons expands 3-fold (Figure 3C), similar to other cortical regions (Bandeira et al., 2009; Ray and Brecht, 2016). This expansion of cortical thickness may confound the analysis of laminar position of dLn neurons. As pointed out previously, direct evidence for any of these migratory patterns is lacking (Bruguier et al., 2020), and further cell lineage tracing studies are necessary to understand the changing patterns of Nurr1 expression during development. Such studies should also take into account the complex developmental patterns of other cortical neurons (Saito et al., 2019).

As discussed by Puelles (2014), differences in pin-pointing the birth dates of claustrum neurons may stem from methodological issues. Bayer and Altman (1991) and earlier studies used radioactive [H³]Thymidine given at a single or multiple time points. In [H³]Thymidine studies, cell types are mainly inferred by their anatomical location. Arimatsu (1994) and Arimatsu et al. (2003) used antibodies against Latexin or Nurr1 in combination with the thymidine analog 5-bromo-2′-deoxyuridine (BrdU) to uncover the birthdates of claustrum and Nurr1/Lxn positive neurons in the lateral cortex. However, BrdU detection requires aggressive denaturation which may impact co-labeling with antibodies or in situ hybridization (Miller and Nowakowski, 1988; Chehrehasa et al., 2009). Administration of BrdU may also interfere with neuroblast proliferation and promote apoptotic cell death (Rodríguez-Vázquez and Martí, 2021). Therefore, we decided to use the more recently developed thymidine analog EdU, which does not require DNA denaturation and has been shown to be a specific and reproducible labeling tool to study cell proliferation in the adult central nervous system (Zeng et al., 2010). We also considered FlashTag labeling which allows for birth dating in a more precisely defined time window (Govindan et al., 2018). However, FlashTag labeling, so far, appears restricted to investigating dorsal cortex progenitors (Yoshinaga et al., 2021). One drawback of EdU is its even higher toxicity than BrdU at comparable dosing regimes (Ross et al., 2011; Andersen et al., 2013). However, since EdU is significantly more easily detected, it was possible to administer EdU at a fraction of the dose of BrdU in previous studies (Figure 1B; Arimatsu et al., 2003). We found that even at the lowest injection dose tested (10 mg/kg), EdU remained readily detectable using our staining protocol. The overall EdU labeling pattern in general agrees with previous birth dating studies. For example, in the cortex there is a clear inside-out EdU labeling pattern and lateral parts of the cortex show precocious development compared with dorsal regions (Figures 1C,D; Hicks and D’Amato, 1968; Smart and Smart, 1977; Saito et al., 2019). Further studies combining, for example, BrdU and EdU as shown by Harris et al. (2018), but using injections on different days, will enable direct comparisons of cells born at different time points. While we focus on EdU labeling in the claustrum and related cell types, our data may be of use to researchers interested in the birth dating of other brain regions, therefore we provided high resolution versions of Figures 4–9 online.¹

Previous birth dating studies of claustrum and DEn neurons have led to conflicting results (see Puelles, 2014, 2017; Bruguier et al., 2020 for reviews). Some earlier studies had suggested that rat claustrum neurons are born before E12.5 (Bisconte and Marty, 1975; Valverde and Santacana, 1994), however, our data indicate that the majority of all claustrum, DEn, and related Nurr1 positive neurons are born after E12.5 (Figure 10). We think that using reference marker, such as Nurr1, greatly expedites the assignment of birth dates to defined neuronal populations. Presumably, the lack of such a reference marker explains the major differences in the interpretation of the data in Bisconte and Marty (1975) and Valverde and Santacana (1994). Further differences may stem from the use of thin sections in [H³]thymidine studies (typically 5 μm), whereas we use relatively thick sections (50 μm), and differences in defining embryonic days. Our claustrum birth dating results agree to a larger extent with the findings of Bayer and Altman (1991); Arimatsu (1994), and Arimatsu et al. (2003). However, it should be noted that Bayer and Altman (1991) define the first day of embryonic development as E1, while Arimatsu (1994) and Arimatsu et al. (2003) define it as E0, whereas we define it as E0.5. For example, Arimatsu (1994) observe that the majority of DEn neurons are born on E13, we find that they are mainly born on E13.5 to E14.5, while Bayer and Altman (1991) show DEn birth dates from E14 to E15. In addition, we further subdivided DEn into three regions. vDEn has lower Nurr1 expression levels and shows the earliest birth dates (mainly on E13.5), whereas dDEn and pDEn have higher Nurr1 expression and later birth dates (on E13.5 and E14.5). Medial to the vDEn, we observed a group of cells called reservoir or paraclaustral reservoir (Bayer and Altman, 1991; Maciejewska et al., 1999) or more recently, bed nucleus of the external capsule (BEC) (Puelles, 2014; Puelles et al., 2016b). Interestingly, these cells show distinctly later birth dates (E15.5-E17.5) than the neighboring DEn (Figures 7–10). These Nurr1 negative neurons are either thought to migrate into olfactory cortex regions within the first few days after birth (Maciejewska et al., 1999) or constitute a more stable nucleus (Puelles, 2014; Puelles et al., 2016b). In both cases, they are unlikely to be related to claustrum neurons. When considering the adjacent cortical regions, we find that the olfactory cortex cells are produced from E13.5 to E15.5 and insular cortex cells are generated mainly on E14.5 to E15.5, indicating that DEn and claustrum develop jointly with their surrounding cortical regions (Puelles et al., 2016a). In contrast, the paraclaustral reservoir/BEC cells are born significantly later than the adjacent olfactory cortex neurons. Again, when we divide the claustrum into the vCL and dCL regions, they show largely similar birth dating patterns. Interestingly, the dLn Nurr1 positive neurons are generated largely on the same days as claustrum neurons whereas sLn neurons are generated significantly later (Figures 11, 12). Together with the strong reduction in Nurr1 expression during early postnatal development, this indicates that sLn neurons constitute a distinct neuronal type, whereas dLn neurons are more closely related to other claustrum neurons, in agreement with their claustrum-like gene expression profile (Bruguier et al., 2020; Peng et al., 2021).

Bayer and Altman (1991) describe a ventral to dorsal neurogenetic gradient in the DEn and a posterior to anterior neurogenetic gradient in the claustrum. In general, our data also support the notion of neurogenetic gradients in CL and DEn, but we find that both types of neurogenetic gradients are present in DEn and claustrum (Figures 10, 13). Thus, Nurr1 positive neurons are born earlier in ventral and posterior regions and later in dorsal and anterior regions within the major claustrum and DEn subregions. In contrast to Arimatsu (1994) and Arimatsu et al. (2003), we also identify ventral to dorsal neurogenetic gradients in lateral cortical Nurr1 positive dLn and sLn neurons (Figure 11). Overall, the prenatal development of Nurr1 expression follows the birth dating pattern, although it remains to be determined how long after neuronal production Nurr1 expression is induced. For example, claustrum and DEn neurons can be differentiated by their birth dates between E13.5 and E15.5, but based on Nurr1 staining alone, these subdivision are only recognizable after E17.5. Thus, experiments combining EdU injections with Nurr1 staining during embryonic development could provide clues about the relative timing of neuron production and induction of Nurr1 expression. More detailed fate-mapping studies are required to determine the origin and trajectory of claustrum and Nurr1 positive neurons in the lateral cortex.