In search of common developmental and evolutionary origin of the claustrum and subplate

The human claustrum, a major hub of widespread neocortical connections, is a thin, bilateral sheet of gray matter located between the insular cortex and the striatum. The subplate is a largely transient cortical structure that contains some of the earliest generated neurons of the cerebral cortex and has important developmental functions to establish intra‐ and extracortical connections. In human and macaque some subplate cells undergo regulated cell death, but some remain as interstitial white matter cells. In mouse and rat brains a compact layer is formed, Layer 6b, and it remains underneath the cortex, adjacent to the white matter. Whether Layer 6b in rodents is homologous to primate subplate or interstitial white matter cells is still debated. Gene expression patterns, such as those of Nurr1/Nr4a2, have suggested that the rodent subplate and the persistent subplate cells in Layer 6b and the claustrum might have similar origins. Moreover, the birthdates of the claustrum and Layer 6b are similarly precocious in mice. These observations prompted our speculations on the common developmental and evolutionary origin of the claustrum and the subplate. Here we systematically compare the currently available data on cytoarchitecture, evolutionary origin, gene expression, cell types, birthdates, neurogenesis, lineage and migration, circuit connectivity, and cell death of the neurons that contribute to the claustrum and subplate. Based on their similarities and differences we propose a partially common early evolutionary origin of the cells that become claustrum and subplate, a likely scenario that is shared in these cell populations across all amniotes.

The claustrum has extremely widespread reciprocal connections with the cortex, which prompted ideas about its potential function in human perception and delusional phenomena (Patru & Reser, 2015).
The subplate is a transient compartment of the developing cerebral cortex, with migrating neurons that populate the overlaying cortical plate after passing through cortical afferent and efferent fibers (Judaš, Sedmak, & Kostovi c, 2013;Kostovi c & Judas, 2010). The subplate contains some of the earliest born neurons of the cerebral cortex (Angevine & Sidman, 1961;Bayer & Altman, 1990;Duque, Krsnik, Kostovi c, & Rakic, 2016;Kostovi c et al., 2014;Price, Aslam, Tasker, & Gillies, 1997;Rakic, 1974;Rickmann, Chronwall, & Wolff, 1977). Subplate cells have important developmental functions that establish intra-and extracortical connections through transient connectivity that will eventually be dismantled at later stages (Allendoerfer & Shatz, 1994;Kanold & Luhmann, 2010). In most species assessed to date, some subplate cells undergo regulated cell death and their transient connectivity is demolished, although it remains a subject of debate as to whether this occurs in the majority or just of few of the cells. In humans and macaque monkeys, some cells remain as interstitial white matter cells. In mice and rats a compact layer, Layer 6b, is formed from the earliest born subplate neurons and remains at the base of the cortex adjacent to the white matter into adulthood (Duque et al., 2016;Hoerder-Suabedissen & Molnár, 2012Price et al., 1997). That Layer 6b in rodents is homologous to the primate subplate and interstitial white matter cells is a likely possibility; although it warrants further investigation, for the purposes of this review we shall assume it as a given.
Undisputed is the fact that the transient subplate compartment is considerably enlarged in primates as compared to other mammals (Kostovi c & Judas, 2010;Wang et al., 2011). It has been argued that the subplate and its postnatal remnant serve as the main area for the evolution of developmental shifts, such as increased brain size, neuron number, connectivity, and shaping of circuits in response to their social environment. Judaš et al. (2013) also proposed that a more complex and protracted development, including intricate transient circuits, might have contributed to more sophisticated cortical organization and consequently more advanced cognitive functions, such as language, self-awareness, and cognition.
In the adult mouse neocortex, a thin sheet of neurons, Layer 6b, appears beneath Layer 6a in the cortical plate, located immediately superficial to the white matter, and has a continuous transition with the deepest part of the claustrum, with which it shares various gene expression patterns Puelles, 2014;Wang et al., 2011;Watakabe, Ohsawa, Ichinohe, Rockland, & Yamamori, 2014; see Section 1.5). In addition to overlapping gene expression patterns, the claustrum and Layer 6b share other developmental features. Notably, neurons in these structures are born earlier than all other cortical pyramidal neurons . The development of the mutual connectivity of the claustrum, dorsal endopiriform nucleus (EPd), lateral amygdala, and insular/piriform cortices is highly dependent on the precise choreography of early axonal guidance mechanisms, which include the formation of reciprocal thalamocortical projections at the pallial-subpallial boundary (González-Arnay, González-Gómez, & Meyer, 2017;Molnár, 1998;Molnár, Garel, López-Bendito, Maness, & Price, 2012). The relatively early histogenesis of all these structures may be highly relevant for our understanding of the development and evolution of forebrain circuits. Here, we appraise the comparative data on cytoarchitecture, evolutionary origin, gene expression, cell types, birthdates, neurogenesis, lineage, migration, connectivity, and cell death of the neurons that contribute to the claustrum and subplate. While we have made every effort to address each topic individually, often multiple strands of evidence are required to understand historic viewpoints or current conclusions. In this review we note the marked similarities between claustrum and subplate neurons and raise the possibility of a partially common developmental and evolutionary origin.

| Cytoarchitecture of the claustrum
The claustro-insular complex forms the lateral pallium (LPall) radial domain (Puelles, 2014), with its immediate dorsal neighbor being the dorsal pallium (DPall), where layers 1-6a/b of the neocortex can be found (6b being equivalent to subplate in rodents). In most mammals, including marsupials and most eutherian species, the claustrum is located near the insular cortex, which in humans and other primates lies beneath the Sylvian fissure between the temporal and parietal/ frontal cortices (Figure 1). While in humans and other primates the extreme capsule is easily detected between the clastrum and insular cortex, in other mammals, such as rodents, microchiropterans, and small marsupials, these fibers are less prominent or even absent and the claustrum lies directly under the deep layers of the insular cortex.
The claustrum (Cl) and EPd together are sometimes referred to as the claustral complex (Puelles, 2014). The EPd lies beneath the piriform cortex within the VPall. A distinct bundle of VPall radial glia crosses the EPd to reach the piriform cortex. At the same level there is a separate ventral endopiriform nucleus next to the putamen, which originates from the VPall and is unrelated to the claustrum. Both the claustrum and the EPd derive from the lateral pallium (Puelles, 2014;Watson & Puelles, 2017), and serve as major hubs of neocortical and limbic circuits, respectively. In general, while the shape and size of the claustrum is variable across mammals, its size increases in proportion with the neocortical volume (Baizer, Sherwood, Noonan, & Hof, 2014;Druga, 2014; Kowia nsk, Dziewiątkowski, Kowia nska, & Mory s, 1999; Figure 1).

| Cytoarchitecture of the subplate/Layer 6b
The subplate is a largely transient cortical structure that contains some of the earliest generated neurons of the cerebral cortex (Allendoerfer & Shatz, 1994;Kostovi c & Rakic, 1990). The subplate can be identified in primate brains by its expression of proteoglycans, with cortical afferents passing through this region and transient and migratory cell populations (Judaš et al., 2013;Kostovi c & Judas, 2010). The lower subplate boundary is not as easy to identify in the primate cortex as it is in the rodent cortex, and in primates the subplate also lacks a clear boundary with the intermediate zone. In the rat and the mouse the subplate appears as a thin, compact cell layer which is increasingly separated from the cortical plate at embryonic (E) stages E14-16 according to an antero-posterior and ventrodorsal gradient (Figure 3b,e;Oeschger et al., 2011). In the putative primary visual cortex of the macaque monkey and human, the subplate divides into upper and lower compartments with a thin, dense, transient cell layer forming at their boundary (Hoerder-Suabedissen & Molnár, 2015;Kostovi c & Rakic, 1990). Such transient compartmentalization has not been observed in other species to date. The subplate is also a compartment in which thalamocortical afferents wait and form early synapses, before growing into the overlying cortex (Kostovi c & Rakic, 1990).
The constituent cells of the transient subplate compartment may survive at least in part as interstitial white matter neurons in monkey and human brains (Duque et al., 2016). In postnatal and adult rodent brains the location of the former subplate is occupied by a densely packed band of cells-Layer 6b-that is distinct from the underlying white matter and separated from the overlying cortex by a thin, cell sparse region.

| Evolutionary comparison of the claustrum
The claustrum is an evolutionarily ancient structure, with a putative claustrum having been described in mammals, birds, and reptiles.
Despite its presence in most mammals studied to date, the existence of a claustrum in monotremes (i.e., egg-laying prototherian mammals) has been long debated. Abbie (1940) examined the brains of platypus and echidna but failed to find a cytoarchitectonically distinct claustrum, and concluded that therefore no associated insular cortex was present. Butler and colleagues argued that the apparent lack of the claustrum in monotremes posed challenges to the understanding of pallial evolution and argued that this might just be a cytoarchitectonic absence, rather than the true absence of the particular cell group Butler, Molnár, & Manger, 2002;. Subsequently Ashwell, Hardman, and Paxinos (2004) reported the presence of both the claustrum and EPd in platypus and echidnas based on the examination of cytoarchitectural, F I G U R E 1 Nissl-stained coronal brain sections and phylogenetic relationships of representative mammalian species (western gray kangaroo, rock hyrax, giant anteater, spotted hyena, and human). Insets on the right were taken from the right hemisphere from the region of the rhinal fissure (rf) near the insular cortex (Ins). The claustrum (Cl) is present in most therians immediately adjacent to the insular cortex and forming a cytoarchitectonic complex with the dorsal endopiriform nucleus (EPd), which lies deep to the piriform cortex (Pir). Scale bar: 3000 μm. Images taken from brainmuseum.org [Color figure can be viewed at wileyonlinelibrary.com] histochemical, and fiber stained material. Puelles (2014) agreed with the evidence that there is a cryptic claustrum in monotremes, and commented that this structure is difficult to distinguish from the insular cortex using cytoarchitectonic or myeloarchitectonic methods, as has been reported in microchiropteran bats (Humphrey, 1936;see references in Puelles, 2014), tree shrews (Tigges & Totada, 1969) and other therian mammals. A full unequivocal elucidation of the monotreme claustral complex is yet to be achieved, and would likely involve complementing comparative analysis of cytoarchitecture (see Figures 1 and 2) with enriched or specific gene expressions (see Section 1.5). In Figure 2 we present high-resolution Nissl images of the platypus brain in coronal and sagittal views indicating tentatively the putative claustrum ( Figure 2). Figure 2 also clearly indicates the EPd and piriform cortex (Pir). In the sagittal plane, the claustrum seems to comprise cells scattered within the white matter above the external capsule and possibly intermingled with an extreme capsule. According to Ashwell F I G U R E 2 Nissl-stained coronal and sagittal sections of the adult platypus (Ornithorhynchus anatinus, Monotremata) brain. (a-f) coronal sections showing the location of the piriform cortex (Pir), rhinal fissure (rf), insular cortex (Ins), anterior commissure (ac), and external capsule (ec). (c, d, f) are higher magnification images of the square area highlighted in (a), (c), and (e), respectively. The dorsal endopiriform nucleus (EPd) lies deep to piriform cortex Layers 1-3 (Pir1-3 in c and d), and the putative claustrum (Cl) includes neuron-sized cells (arrows in f), possibly intermingled with white matter of an extreme capsule (extc  (Ashwell et al., 2004). Whether this evidence corresponds to a true claustrum is unclear and would require additional confirmation, such as the expected expression of claustrum-specific marker genes discussed below. Interestingly, a recent study from Norimoto et al. (2020) used single-cell transcriptomics to identify a putative claustrum in both the Australian bearded dragon and a species of turtle. While this study has yet to be replicated, it suggests that the claustrum may even have been present in the common ancestor of all mammals and reptiles.
The ancestral embryological origins, and potential homology of the mammalian claustrum with that of other vertebrates is currently not fully elucidated. Puelles (2014) proposed that a claustro-insular homolog was already present in ancestors of all modern amniotes, such that in chicken it corresponds to the lateropallial mesopallium of the dorsal ventricular ridge (DVR). Puelles assigned the embryological origin of both the claustrum and insula to the lateral pallium (first conceived for the mouse in Puelles, 2014). This implies that this complex should not be considered neocortical, as it does not originate from DPall. A similar opinion was held by Holmgren (1925), who considered the claustrum as a pallial structure not derived from the neocortex.
Butler and Molnár postulated the collopallial field hypothesis, which examined the differences in the claustroamygdalar formation between birds, reptiles, and mammals . The hypothesis held that in mammals, the thalamo-recipient collopallium differentiates into deep (claustroamygdalar) and superficial (neocortical) components, whereas in sauropsids, and perhaps in platypus, this split may not occur. The original collopallial field hypothesis was based on a handful of gene expression patterns , and it was formulated before the modern tetrapartite pallium model that is now commonly used. The modern concept of VPall (olfactory), LPall (claustro-insular), and DPall (neocortical) pallial sectors (Puelles, 2014(Puelles, , 2017 seems to have rendered obsolete these notions articulated in Puelles et al. (2000) and Molnár and Butler (2002). Altogether, the general presence of the claustrum in all extant mammalian taxa (excluding the uncertainty in monotremes), together with shared histochemical, molecular, cytoarchitectural, and hodological features, suggests that the claustrum was already present in the now extinct pan-mammalian common ancestor Puelles, 2017;Suárez et al., 2018).

| Evolutionary origin of the subplate
The six-layered cerebral neocortex and the claustrum-insula complex have been widely reported in therians, marsupials, and placentals, but the existence of a cortical subplate in marsupials has been the subject of controversy Pearce, James, & Mark, 2000).
Subplate markers identified in the mouse are expressed in the opossum Monodelphis domestica, but are much more dispersed than in eutherians Puzzolo & Mallamaci, 2010;Wang et al., 2011). Moreover, thalamic cortical projections, thought to also represent subplate afferents, have a different spatiotemporal pattern in their distribution and arrival to the cortex in marsupials as compared to rodents or carnivores (Molnár, Knott, Blakemore, & Saunders, 1998). This pattern is compatible with the distribution of the early generated and subplate enriched gene expression in M. domestica Montiel et al., 2011). Subplate enriched genes are also expressed in the developing reptilian and avian cortices .  (Aboitiz, Montiel, & García, 2005;Marin-Padilla, 1978), (b) the derived hypothesis states that the subplate is a novel character exclusive to mammals with complex and larger brains, whose emergence supported the development of expanded cortico-cortical connectivity (Kostovi c & Rakic, 1990;Molnár et al., 2006;Supèr & Uylings, 2001), and (c) the dual origin hypothesis suggests that there are both ancestral elements (so-called "presubplate" cells) (Kostovi c & Rakic, 1990;Meyer, 2007;Meyer, Castro, Soria, & Fairén, 2000) and newly emerged cell populations in the mammalian subplate, with additional populations of subplate cells being added to the embryonic subplate as the neocortex became more complex (Aboitiz, 1999;Kennedy & Dehay, 2012;Meyer, 2007;Montiel et al., 2011;Suárez-Solá et al., 2009;Wang et al., 2011). These hypotheses could be evaluated by further researching birth dates, tracking cell origin and lineage, and examining somatodendritic morphology, physiological properties, and single cell transcriptomics, as well as markers for differential properties among ancestral and newly developed subplate cells in various species (Hoerder-Suabedissen & Molnár, 2015). Understanding how the subplate has been altered in distinct mammalian lineages would help to understand which of these hypotheses is correct. The available evidence, however, suggests that the human subplate contains an increased number of both ancestral and derived subplate neurons (Judaš et al., 2013;Meyer, 2007;Suárez-Solá et al., 2009). Moreover, any such analysis is further complicated by the fact that the subplate compartment in development contains many radially and tangentially migrating cells that are just temporarily passing through.

| Gene expression studies in the claustrum and cerebral cortex
The upper and deeper layers of the cerebral cortex (layers 2/3 vs. 5/6 in neocortex) have distinct neurogenic programs regulated by differential transcriptional networks in mammals (Britanova et al., 2008;Molyneaux et al., 2009;Molyneaux, Arlotta, Menezes, & Macklis, 2007;. Some of these programs specify the deeper layer neurons with extracortical projections, the upper layer neurons with mostly intracortical connectivity, while the granular layer 4 in between them receives most of the thalamic input. Importantly, the differential expression of supragranular (upper layers) and infragranular (deeper layers) cortical genes Satb2 and Fezf2 in the presumptive claustrum and amygdaloid nuclei of mice at E15.5, suggest that they may have different developmental origins (Figure 3a; Paolino et al., 2018). Interestingly, claustrum and subplate/Layer 6b (L6b) neurons also express canonical markers of cortico-cortical neuronal types, such as the callosal determinant including subplate and Layer 6b (Belgard et al., 2011;Hoerder-Suabedissen et al., 2009;Oeschger et al., 2011). The claustrum expresses many of the same markers that are enriched in the subplate Wang et al., 2011). However, not all genes that are subplate-expressed in the mouse are also present in the mouse claustrum. There are important species differences in gene expression, even between closely related species such as rats and mice, both in terms of whether a particular gene is a "subplate marker" as well as when comparing subplate and claustrum gene expression Wang et al., 2011). In addition to the cortical subplate, Nr4a2/Nurr1 is selectively expressed in the claustrum and in layers 5 and 6 of the parietal neocortex in both mice and rats , as well as in macaque monkeys (Watakabe et al., 2014).
Another subplate specific gene, Complexin 3 (Cplx3) is exclusively expressed in the subplate of both mouse and rat (Hoerder-Suabedissen et al., 2009;Wang et al., 2011). Mouse connective tissue growth factor (Ctgf ) only labels the subplate component in neocortex and the deep part of the claustrum and endopiriform nucleus complex (Hoerder-Suabedissen et al., 2009). DOPA decarboxylase (Ddc) is expressed in the mouse subplate, while in rat, it is found in the EPd . Transmembrane Protein 163 (Tmem163) is another subplate-specific gene in mouse and rat dorsal cortex, but additional expression in the basomedial amygdala, bed nucleus of the stria terminalis, and medial amygdala is only observed in the mouse . Moreover, Monooxygenase DBH Like 1 (Moxd1) and thyrotropin-releasing hormone (Trh) are only expressed in the subplate of mouse but not in rat , and differ in their non-subplate expression between these two species. Moxd1 is expressed in endopiriform nucleus, central amygdala, while Trh is expressed in the bed nucleus of stria terminalis in rat, but not mouse brains . Mathur et al. (2009) described G-protein gamma2 subunit (Gng2) as a novel protein that is selectively expressed in claustrum only at striatal, but not in frontal levels.

T A B L E 1
Genes expressed in the claustral complex (claustrum and endopiriform nucleus) of adult mice assessed for co-expression in neocortical layers (1) Claustrum (2) Claustrum + EnP (3) Claustrum + EnP + L6b (4) Claustrum + EnP + lateral deep layers (5) Claustrum + EnP + lateral deep layers w/o L6b (6) Claustrum + EnP + deep layers Thsd7b Col11a1 Tmem163 Trp53i11 Usp46 Note: 44 genes with enriched claustrum expression were identified using the Allen Brain Atlas fine structure search tool. Of these, three were discarded as they showed weak and/or uniform expression. Each of the remaining 41 genes was placed into the best fitting single category (1-7). Genes in bold were independently identified as "subplate/Layer 6b markers" in one of our previous microarray or RNAseq experiments (Hoerder-Suabedissen et al., 2013). Group assignment based on RNA in situ data from the Allen Brain Atlas: Data Portal. 1.6 | Claustrum-and Layer 6b-enriched genes can be classified into distinct sub-groups according to their patterns of expression in other structures To systematically explore the similarities and differences in the patterns of gene expression between subplate/Layer 6b and claustrum in the adult mouse, we identified claustrum-enriched genes using the "fine structure search" tool on the Allen Brain Atlas: Data Portal (ABADP; https://www.nitrc.org/projects/abaportal/), a manually curated list of up to 50 genes with enhanced gene expression in different small brain structures. We also identified subplate/Layer 6b-enriched genes by using the "fine structure search" tool on the ABADP. The former group T A B L E 2 We examined genes that were preferentially expressed in Layer 6b based on the Allen Brain "fine structure search tool" and examined in situ hybridization data from the Allen Brain Atlas: Data Portal for co-expression in other brain structures (1) L6b (with or without expression in some other cortical layers or striatum) (2) L6b and dEnP (3)  Note: Of the 48 selected subplate/Layer 6b-enriched genes, 14 had to be discarded, as they were not L6b specific on examining the ISH signal. The remaining 34 genes were placed into the best fitting single category (1-9) based on the pattern of co-expression in additional structures. Layer 6b enriched genes were also expressed in claustrum (see Groups 3, 6, 7, 8, and 9), but there were several other expression patterns of Layer 6b genes without claustral expression as well (see Groups 1, 2, 4, and 5). Genes in bold are also present in Table 1 of "claustrum enriched" genes.
of genes was additionally annotated for whether the genes were identified as subplate-enriched in one of our previous microarray surveys (see raw data in Belgard et al., 2011;Hoerder-Suabedissen et al., 2009Oeschger et al., 2011), whereas the latter table was additionally annotated to highlight the subset of genes identified by the "fine structure search" tool as both claustrum and Layer 6b enriched.
Then, using the in situ hybridization data available on the Allen Institute we identified the structures these genes were expressed in, focusing specifically on the cortical layers, EPd, claustrum, and striatum. We then grouped these genes according to their common co-expression patterns. We found that the claustrum-enriched genes could be classified into one of at least seven different patterns of expression (Table 1).
Similarly, subplate/Layer 6b-enriched genes could be classified into at least nine common co-labeling patterns, some of which were distinct from the claustral co-expression patterns ( Table 2). F I G U R E 4 Schematic sagittal sections of E15.5 mice processed for in situ hybridization for selected markers of the adult claustrum. Nr4a2, also known as Nurr1, is one of the strongest and most exclusive markers of the lateral pallium (LPall) and putative claustrum. Note a continuum of labeled neurons into the cortical subplate is present in most cases, some stronger (e.g., Nr4a2 and Cdh10), and others are much weaker if at all (Gnb4). Scale bar: 300 μm. Images taken from the Allen Brain Atlas Data Portal [Color figure can be viewed at wileyonlinelibrary.com] In situ hybridization signal is not necessarily absent/present in particular brain regions, but may vary in signal strength across different cell groups. Thus, genes were assigned to the single best-fitting category, but weaker or sparser expression was often present in other regions.
Interestingly, the largest group of gene co-expression was in Group 3, the neocortical Layer 6b (Table 1) Conversely, we assessed the 46 genes identified as "Layer 6b enriched" by the Allen Brain Atlas fine structure search tool for coexpression in the claustral complex. Of these, 12 genes were F I G U R E 5 Examples for genes that were enriched in in Layer 6b and endopiriform nucleus and not claustrum (Ctgf, Nxph4, upper two rows (a)-(e) and (f)-(j)) and for genes with expression in claustrum and dorsal endopiriform nucleus but not within Layer 6b expression (Gnb4, Nmb, lower two rows (k-o) and (p-t)). This figure shows sagittal sections of embryonic (E) 15.5, 18.5, postnatal day (P) 4 and 56 (Adult) and an additional coronal section (right column) of the adult mouse brains after in situ hybridization against Ctgf, Nxph4 (upper two rows) and (Gnb4, Nmb, lower two rows). Expression of Ctgf and Nxph4 is evident in the subplate and subplate-derivative Layer 6b (arrows) together with the expression in the endopiriform nucleus (double arrowheads), but not in claustrum, where there is a clear gap in expression. Expression of Gnb4 and Nmb is evident in the claustrum (single arrows) and dorsal endopiriform nucleus (double arrowheads). Additional genes that share similar expression patterns to Ctgf and Nxph4 are listed in Group 2 of Table 2; patterns similar to Gnb4, Nmb are listed in Groups 1 and 4 of Table 1. Scale bar: 500 μm. Images were taken from the Allen Brain Atlas Data Portal [Color figure can be viewed at wileyonlinelibrary.com] discarded from further analysis as the Layer 6b-enriched expression could not be verified due to weak or uniform signal strength. The remaining 34 genes were classified into nine distinct co-expression patterns, based on expression in claustrum, dorsal endopiriform cortex, other neocortical layers, and striatum. As before, genes were assigned to the single best-fitting category, but weaker or sparser expression was often present in other regions. The majority of Layer 6b enriched genes were co-expressed in the claustral complex, but a sizeable fraction of those were not expressed in the claustrum, but just in the endopiriform nucleus. Six genes showed exclusive neocortical expression (Group 1 in Table 2), 10 genes were expressed in Layer 6b and endopiriform nucleus, but not claustrum (Group 2 in Table 2, see examples of Ctgf and Nxph4 in Figure 5.). Seven genes showed similar expression strength and density across all of L6b, claustrum, and EPd (Group 3, Table 2), essentially giving a continuous band of expression from dorsal Layer 6b to endopiriform nucleus. Some of the genes are not only expressed in 6b, but also in 6a and even 5 (Arimatsu et al., 2003;Hoerder-Suabedissen et al., 2009Puelles, 2014) and they showed areal differences in expression. Insofar as some of the Layer 6b enriched genes show a continuation with the claustrum, whereas others do not, as described by Wang et al. (2011), it may be suggested that possibly some, but not all, claustrum cells might share a common origin with subplate cells.
Overall, the most prominent co-expression pattern in the adult mouse brain was Layer 6b with endopiriform nucleus (without claustrum), closely followed by continuous L6b to endopiriform nucleus (including claustrum) expression. Only a small group of genes were expressed in deep layers of cortex and the claustral complex. Overall, four genes from the "Layer 6b enriched" Table 2 are also present in Table 1 of "claustrum enriched" genes. Nurr1/Nr4a2 was not included among the genes with "Layer 6b enriched" expression or in the "claustrum enriched expression" in the Allen Brain Atlas fine structure search tool lists. Nonetheless, the partial overlap of gene expression between Layer 6b and claustrum can be seen as partial validation of the gene-based hypothesis of Puelles (2014), which only referred to a significant, but not total, claustral contribution to the subplate, as well as to the deep layers of neighboring neocortex, as previously described by Arimatsu et al. (2003). We focused on adult gene expression, because this data set is the most comprehensive in the Allen Brain Atlas in situ hybridization image database. As

| Cell types of subplate and claustrum
The claustrum is composed of two primary cell types, previously described as Type I and Type II neurons (Braak & Braak, 1982;Crick & Koch, 2005). The claustrum contains spiny, glutamatergic excitatory neurons as well as GABAergic inhibitory neurons in roughly the same proportions as the cerebral cortex (Mathur, 2014). The excitatory neurons of the claustrum are enriched in several genetic markers, including Gng2, Gnb4, and Slc17a6 (Hur & Zaborsky, 2005;Mathur, 2014).

| Putative common developmental origin of claustrum and insular cortex
There was a long-standing controversy as to whether the claustrum was formed by a duplication of the suprajacent insular cortex (Holl, 1899) or the piriform cortex (Smith, 1910), and attempts were made to link its development to cytoarchitectonic differentiation (Mathur, 2014). Sonntag and Woollard (1925) suggested that the extreme capsule separates the deepest layer of the insular cortex from the claustrum (Mathur, 2014). Brodmann (1909) considered the claustrum as a sublayer of the insular cortex split off from this cortex to generate a seventh layer. Rose (1928) deduced that in mammals with or without the extreme capsule the claustrum either represented the innermost extension of insular cortex Layer 6, or formed an independent cortical layer, Layer 8 (Mathur, 2014). However, Ramon and Cajal (1902) pointed out that claustral neurons never had apical dendrites pointing into insular layer 1; considering this an argument against the idea that claustrum was an insular layer, suggesting a hypopallial nucleus origin (Puelles, 2014). Watakabe et al. (2014) also noted that the dendrites of insular and claustral neurons do not cross the border of the two brain regions and dendrites of claustral neurons do not invade the overlying insular territory. There is also scant evidence of columnar connectivity between the claustrum and insular cortex. Modern cortical experts accordingly identified the claustrum as a separate pallial nucleus, which develops precociously within the same pallial progenitor domain as the later-born insular population (Puelles, 2014).
The common origin of claustrum and insular cortex was recently suggested on the basis of mapping of Gng2 and NtrG2 markers in the adult human brain (Pirone et al., 2012). Puelles (2014) argued that the development of the claustrum is associated with the subsequently developing insular primordium of the lateral pallium via a shared field of radial glia. The claustral population is born first and takes its position at the thin and superficial LPall mantle layer of E12.5 mouse embryos (in the mouse [Hinds & Angevine, 1965;Smart & Smart, 1977; in the rat [Bisconte & Marty, 1975;Valverde & Santacana, 1994;Valverde, Facal-Valverde, Santacana, & Heredia, 1989;Valverde, Lopez-Mascaraque, Santacana, & De Carlos, 1995]; reviewed in Puelles [2014]). Insular cells subsequently migrate radially (inside-out) through the claustrum, similarly as standard supragranular cortical cells migrate through infragranular cortical cells, thus occupying progressively more superficial positions (Puelles, 2014). This neurogenic pattern suggests that the development of the insula is linked to the claustrum. In later-stage embryos of carnivores and primates the claustrum separates cytoarchitectonically from the insular cortex primordium with the development of the extreme capsule. However, in rodents, bats, and other cases the claustrum remains embedded within the insula, though its cells keep their molecular distinctions from insular neuronal populations (Narkiewicz & Mamos, 1990;Puelles, 2014;Puelles, Alonso, García-Calero, & Martínez-de-la-Torre, 2019).
Additional evidence suggests the existence of a pseudolamination within the mouse claustrum, namely distinguishing core and shell components, as suggested by Real et al. (2006), which cannot be related strictly to the origin of neocortical layers (excepting in case of tangential palliopallial migrations) (Binks, Watson, & Puelles, 2019).
However, the existence and the validity of the entire core/shell distinction were debated by Mathur (2014). A distinct pseudolamination nevertheless had been reported by Narkiewicz and Mamos (1990) after careful cytoarchitectonic analysis of a variety of basal mammals (their "laminar" and "principal" claustrum parts), and this pattern was later discussed by Puelles (2014) into "subplate" and "principal" claustrum parts (i.e., the "shell" or "laminar" component is a subplate-like population underneath the principal claustrum, which is continuous with the neocortical subplate). Suárez et al. (2018) similarly reported that the claustrum of marsupials is rich in Nr4a2/Nurr1 at its core and it is surrounded by Ctip2/Bcl11b-positive cells. Similarly, the presence of a true EPd (concept introduced by Loo [1931]) is not cytoarchitectonically clear in humans, in contrast to most other mammalian species (Figures 1 and 2). This apparent absence might also be related to deficient or superficial anatomical analysis due to the use of outdated methods. Moreover, in species with a clear endopiriform nucleus, cytoarchitectonically distinct dorsal and ventral components have been described (Hardman & Ashwell, 2012;Paxinos, Watson, Petrides, Rosa, & Tokuno, 2012), although the exact delimitation of each remains imprecise . Puelles (2014) showed that EPd and the claustral primordium are both Nr4a2/Nurr1 positive.
Based on Nr4a2/Nurr1 in situ hybridization, (Puelles, 2014(Puelles, , 2017Puelles et al., 2016) argued that at least part of the subplate population may arise at the claustro-insular lateral pallium domain (insular mesocortex) and then migrates tangentially dorsalward under the cortical plate of the DPall in the plane of the subplate. In addition to cortical subplate Nr4a2/Nurr1 is selectively expressed in the claustral core nucleus, EPd, and in a separate set of later-born lateral pallium neurons at E14.5 and in Layers 5 and 6 of the parietal neocortex in both mouse and rat  and in macaque monkey (Arimatsu et al., 2003;Watakabe et al., 2014). At E14.5 granular and supragranular elements are just being born and therefore prospective neocortical Layers 5 and 6 populations form the whole cortical plate.
Initially, these granular and supragranular cells are largely restricted to ventral parietal cortex, though they subsequently progressively appear in other more dorsal areas.
Puelles' proposed claustral cell contingents either reside locally (LPall claustral subplate) or might migrate tangentially into the DPall subplate and do not necessarily exclude other subplate populations in the neocortex or cingulate cortex (Puelles, 2014). However, not all genes that are subplate-expressed are also present in the claustrum, and some of these genes have diverse extracortical and pallial expression patterns (see Tables 1 and 2 above) and there are also strong species differences in subplate specific or enriched gene expression even between rat and mouse Wang et al., 2011). The attractive hypothesis on the common origin of claustrum and subplate F I G U R E 6 (a, b) In utero electroporation of EGFP at the approximate locus of ventral pallium (VP) and lateral pallium (LPall) at embryonic day (E) 11 labels (c) mainly piriform cortex (VPall cells), and only isolated cells under the claustrum (Cl). Some subplate neurons were labeled inside the claustrum, but only in low numbers from this particular E11 VPall and LPall electroporation. Claustrum is indicated by Nurr1/Nr4a2 immunofluorescence (red in (c)). Note subpial cells labeled according to their topography as "insular" -Ins-most probably included populations of Cajal-Retzius neurons known to be produced at the VPall at E11.5, which then migrate subpially across the LPall into DPall (Bielle et al., 2005). In contrast, the example for the E12 electroporation was mostly restricted to LPall and less VPall (d), and labeled mostly insular cortex (Ins) with no apparent claustrum labeling. Scale bars 500 μm for (c) and ( (Saito et al., 2019). This is believed to be associated to a considerable expansion of the dorsal cortex with ventral streaming of the earliest generated preplate derived neurons, including the subplate. This result, together with hypothetical lateropallial (Puelles, 2014) and cingulate elements migrating tangentially (Pedraza et al., 2014) corroborates that the subplate represents an environment where pallio-pallial tangential migrations take place (Puelles, 2011).
In utero electroporation of the lateral pallium at E12 labels overlying neurons of the insular cortex but no cells migrating dorsally Selected gene expression patterns suggested the existence of a claustrum homolog in non-mammalian amniotes that is relatively large and contributes to the subpial part of the mesopallial DVR and includes as well a credible EPd homolog cell population in VPall, under piriform cortex Montiel et al., 2011;Norimoto et al., 2020;Puelles, 2017;Puelles et al., 2000Puelles et al., , 2016Wang et al., 2011;Watson & Puelles, 2017). Analysis of the entire transcriptome from selected regions of the adult chick and adult mouse brains only partially supported this notion  The above presented preliminary results (Figures 6 and 7) are not quite sufficient to reach a firm conclusion with respect to the different interpretative possibilities (also see figure 6 in Puelles, 2017), nevertheless they portray a glimpse into the lineage relations that we could further explore with these approaches. Refining the precise location and timing of these electroporation experiments will hold the key to reveal possible relationships between subplate and claustrum lineages.

| Comparisons of connectivity of the claustrum and Layer 6b/subplate
The claustrum has the strongest connectivity in the adult human brain by regional volume , and it is considered a major hub, whereas the subplate has the most extensive precocious intracortical and extracortical connectivity during development which is held to be necessary for the guidance of various sorts of axons during development and to orchestrate intracortical and extracortical connectivity (Allendoerfer & Shatz, 1994;Kostovi c & Rakic, 1990;Molnár & Blakemore, 1995). During cortical development, subplate neurons receive glutamatergic input from thalamocortical axons, as well as GABAergic input from interneurons, and establish reciprocal connections with cortical neurons within the local radial column (Friauf & Shatz, 1991;Higashi, Hioki, Kurotani, Kasim, & Molnár, 2005;Higashi, Molnár, Kurotani, & Toyama, 2002;Molnár, Kurotani, Higashi, Yamamoto, & Toyama, 2003;Piñon, Jethwa, Jacobs, Campagnoni, & Molnár, 2009). Some of these connections are remodeled during development, but in the adult mouse there are extensive connections maintained from Layer 6b to local and distant cortical areas, and to thalamus (Hoerder-Suabedissen et al., 2018).
The ipsilateral claustro-cortical projections in rodents are much stronger than contralateral projections White et al., 2017). F I G U R E 9 Photomicrographs of immunostained orexinergic boutons from the insular cortex (a, d, g), claustrum (b, e, h), and subcortical white matter (c, f, i) from the common chimpanzee (a-c), lar gibbon (d-f), and African wild dog (g-i), demonstrating the presence and relative innervation density of these regions of the telencephalon by the orexinergic system. Note that in all cases the orexinergic boutons are present and are comprised of both large and small boutons, as seen in other mammals (Dell et al., 2015). These boutons were revealed with the antiorexin-A antibody (AB3704, Merck-Millipore; RRID AB_91545; raised against a synthetic peptide corresponding to the c-terminal portion of bovine orexin-A peptide). In all images dorsal is to the top and medial to the left. Scale bar in I = 50 μm and applies to all Orexin is a peptide that is delivered from the lateral hypothalamus via wide projections across the entire central nervous system, including to other local (hypothalamic) neurons that are important for modulating arousal, appetite, and activity of the neuroendocrine functions (Li & de Lecea, 2020). Layer 6b cortical cell populations are the only orexin sensitive cell groups in sensory regions of the cerebral cortex (Bayer et al., 2004).
We have little information on the distribution of orexin receptors in the claustrum, but orexin immunoreactive boutons are present in the insular neocortex, claustrum, and subcortical white matter in common chimpanzee, lar gibbon, and African wild dog (Figure 9), demonstrating the presence and relative innervation density of these regions of the telencephalon by the orexinergic system (Manger, unpublished;Dell et al., 2015).
We hypothesize that the influence of the orexinergic lateral hypothalamic neurons on cortex is mediated through fast neurotransmitters and orexin neuropeptides on Layer 6b neurons in broad cortical areas and perhaps a similar influence is mediated through claustrum that also receives moderately dense orexin immunoreactive projections. We hypothesize that this influence of lateral hypothalamus and the sensitivity of cortex and claustrum is also present in human providing targets for potential pharmacological manipulations. Together, these investiga-  (Luskin & Shatz, 1985a, 1985bMeyer et al., 1992), and rodents (Arias, Baratta, Yu, & Robertson, 2002;Ferrer, Bernet, Soriano, del Rio, & Fonseca, 1990;Price et al., 1997); however, in cats, humans and other primates a higher proportion of subplate neurons may survive as white matter neurons (Kostovi c & Rakic, 1980;Meyer et al., 1992;Valverde et al., 1995;Valverde & Facal-Valverde, 1988

| SUMMARY
Here we have reviewed evidence suggesting extensive similarities between the claustrum and the subplate. They both contain principal projection neurons that are among the earliest born in the developing pallium, as well as later-born GABAergic interneurons that arise from subpallium. There is evidence for the general presence of the claustrum in all extant mammalian taxa, but in monotremes there is some uncertainty about the cytoarchitectonic distinctions of the claustrum.
The existence of a claustrum in all extant mammalian taxa in conjunction with recent evidence of a putative reptilian claustrum raises the possibility that the claustrum may already have been present in the common ancestor of all amniotes.
Almost all of the markers we examined that were expressed in the subplate/Layer 6b were also expressed in the claustral complex, but not all claustrum specific or enriched genes were expressed in subplate/Layer 6b, and vice-versa, suggesting overall similarities, but also some differences. Based on gene expression patterns we found the deep (shell or laminar) part of the claustrum the most related to the subplate. The expression patterns in other cortical layers and structures provide some support to the pallio-pallial claustro-neocortex subplate migration and subpial tangential migrations hypotheses, but also highlight the fact that confirmation of such hypotheses by direct lineage tracing methodologies is required. We presented some preliminary cell lineage studies that identified the origin of claustrum and insula, as well as the EPd, in the lateral pallium of the mouse and chick. However, to date there is no conclusive answer about the question of whether the lateral pallium produces a labeled subset of subplate cells which invades the dorsal neocortex, mixing with other subplate cells that are produced locally, or originating elsewhere. The issue is important. In one case, the observed claustrum/subplate similarities are due to shared aspects of patterning, and similar typological and histogenetic properties of adjacent cortical progenitor areas (LPall vs. DPall). In the other case (i.e., claustral cells contributing to neocortical subplate), similarities might be due to migrated claustral elements that conserve some of their original properties. Refining the location and timing of electroporation experiments will hold the key to revealing the possible relationships between subplate and claustrum lineages, while they will also provide ways for defined experimental manipulations. There are numerous similarities between the adult connectivity patters of Layer 6b and claustrum. In particular, both have extensive cortical connections. However, there are differences in the relative strength of their reciprocal connectivity with various other structures in the adult. Similar comparisons at various developmental stages are overdue.
Cell death comparisons between the claustrum and subplate will be required to assess whether these structures have different survival rates. Detailed comparisons through developmental stages will be critical to unravel further similarities and differences in molecular, cellular, and connectivity features between both structures, and will also provide a robust framework for similar comparisons with non-mammalian amniotes. Many questions remain about the claustrum and subpate/ Layer 6b. A better understanding of the evolutionary origins; development; impact in cortical evolution; roles in brain wiring during development; and relation to higher brain functions of these structures will provide important insights. In particular, further research is needed into the neurological diseases and cognitive conditions associated with disruptions of these structures, as well as into potential strategies to design new diagnostic and therapeutic approaches.

CONFLICT OF INTEREST
The authors have no conflicts of interest and that, no interests of any person/organization are affected by the information presented in the present manuscript.

DATA AVAILABILITY STATEMENT
The data that support the findings of Figure 1 are available in Comparative Mammalian Brain Collections at (http://brainmuseum.org). These data were derived from their resources available in the public domain: (http://brainmuseum.org). The data that support the findings of Figure 2 are available in Images taken from BrainMaps.org. These data were derived from their resources available in the public domain: (http:// brainmaps.org). The data that support the findings of Figures 3 and 4 are available in (Allen Brain Atlas Data Portal) at Mouse Brain in Situ Hybridization (ISH) Data (http://help.brain-map.org/display/mousebrain/In +Situ+Hybridization+%28ISH%29+Data). These data were derived from the Allen Brain Atlas Data Portal (ABADP) resources available in the public domain: (http://help.brain-map.org/pages/viewpage.action?pageId= 2424836). The data that support the findings of Figure 5, Tables 1 and 2 were generated from the data that are available in (Allen Brain Atlas Data Portal) at Mouse Brain in Situ Hybridization (ISH) Data (http://help. The data is available from the corresponding author upon reasonable request.