Controlling for activity‐dependent genes and behavioral states is critical for determining brain relationships within and across species

Abstract The genetic profile of vertebrate pallia has long driven debate on homology across distantly related clades. Based on an expression profile of the orphan nuclear receptor NR4A2 in mouse and chicken brains, Puelles et al. (The Journal of Comparative Neurology, 2016, 524, 665–703) concluded that the avian lateral mesopallium is homologous to the mammalian claustrum, and the medial mesopallium homologous to the insula cortex. They argued that their findings contradict conclusions by Jarvis et al. (The Journal of Comparative Neurology, 2013, 521, 3614–3665) and Chen et al. (The Journal of Comparative Neurology, 2013, 521, 3666–3701) that the hyperpallium densocellare is instead a mesopallium cell population, and by Suzuki and Hirata (Frontiers in Neuroanatomy, 2014, 8, 783) that the avian mesopallium is homologous to mammalian cortical layers 2/3. Here, we find that NR4A2 is an activity‐dependent gene and cannot be used to determine brain organization or species relationships without considering behavioral state. Activity‐dependent NR4A2 expression has been previously demonstrated in the rodent brain, with the highest induction occurring within the claustrum, amygdala, deep and superficial cortical layers, and hippocampus. In the zebra finch, we find that NR4A2 is constitutively expressed in the arcopallium, but induced in parts of the mesopallium, and in sparse cells within the hyperpallium, depending on animal stimulus or behavioral state. Basal and induced NR4A2 expression patterns do not discount the previously named avian hyperpallium densocellare as dorsal mesopallium and conflict with proposed homology between the avian mesopallium and mammalian claustrum/insula at the exclusion of other brain regions. Broadly, these findings highlight the importance of controlling for behavioral state and neural activity to genetically define brain cell population relationships within and across species.


| INTRODUCTION
Gene expression profiling of brain regions and cell types have been used as evidence for determining brain region relationships within and across species. One debate using gene expression has been on homologies of cell populations in the pallium across vertebrate species. In search of ventricle subplate cell marker genes that give rise to the mammalian cortical layers and the homologous pallium across vertebrates, Wang et al. (2011) identified the Nuclear Receptor Subfamily 4 Group A Member 2 (NR4A2, also called nuclear receptor related 1, Nurr1). They found high expression levels in embryonic and adult cortical subplate and claustrum cells of mammals, and in the hyperpallium on the dorsal surface of the pallium of chickens. They suggested that the mammalian cortical plate neurons could be homologous to avian hyperpallium neurons. Based on these findings and that of Watakabe, Ohsawa, Ichinohe, Rockland, and Yamamori (2014) on NR4A2 in the claustrum of mammals, Puelles (2014)  1. In mammals, the NR4A2-positive claustrum develops first, followed by some of the NR4A2-positive cells migrating into the insula.
2. In birds, the NR4A2-postive lateral mesopallium is similar to the mammalian claustrum and the medial mesopallium is similar to the mammalian insula, in an outside-in pattern that is opposite to the inside-out pattern of mammals.
3. That lateral part of the avian mesopallium (M), similar to the mammalian claustrum relative to the surrounding brain regions, extends into the avian hyperpallium (H) dorsal to it and nidopallium (N) ventral to it. 4. That the avian NR4A2 pattern in birds contradicts the Jarvis et al., 2013 andChen et al. (2013) mirror image hypothesis of avian brain organization, where the latter two publications renamed the avian hyperpallium densocellare (HD) as dorsal mesopallium (MD). 5. That the NR4A2 patterns in birds compared to mice are inconsistent with the Suzuki & Hirata, (Suzuki & Hirata, 2014) hypothesis that the avian mesopallium is homologous to mammalian cortical layers 2/3.
In follow-up studies and reviews, Watson and Puelles (2017) used the NR4A2 expression pattern to revise their understanding of the relationship of the claustrum to the endopiriform nucleus ventral to it. This included a newly proposed tetrapartite breakdown of the vertebrate pallium: for birds, dorsal as hyperpallium, lateral as mesopallium, ventral as nidopallium, and medial as hippocampus. This view has been debated (Atoji, Sarkar, & Wild, 2018;Puelles, 2017;Wullimann, 2017aWullimann, , 2017b, arguing for a different revision of the tetrapartite organization. Wullimann, 2017a suggests a tetrapartite hypothesis for birds of dorsal as hyperpallium and mesopallium, ventral and lateral as nidopallium, and medial as hippocampus. Wullimann took into consideration the combined findings of Watson and Puelles (2017), Jarvis et al. (2013) and Chen et al. (2013). The latter two studies argued that the gene expression evidence does not support a tetrapartite organization of the avian brain. Puelles (2017) further claimed that the patterns of NR4A2 and another gene, CYP26B, showed that the lateral most edge of the avian mesopallium is strictly the homolog of the mammalian claustrum proper, while the rest of the mesopallium is a field homolog of a combination of the mammalian claustrum and insula cortex. These findings were claimed to further justify support for the tetrapartite organization.
A more recent collaboration between the Puelles and Molnar groups (Bruguier et al., 2020) examined NR4A2 alongside many other genes (40-50 per brain region) from the Allen Institute mouse developmental and adult gene expression Brain Atlases (Lein et al., 2007;Thompson et al., 2014). They found that most claustrum enriched genes were also enriched in cortex layer 6b, but the converse was not found for a number of genes. Further, their preliminary cell lineage tracing experiments found that the newly dividing cells that enter the claustrum or insula from the lateral pallium stay within their respective subdivisions, instead of migrating between these two structures or into layer 6b. Similarly, cells from the avian lateral ventral mesopallium stay within the ventral mesopallium (MV), without migrating dorsally into the lateral hyperpallium or ventrally into the lateral nidopallium. These latter findings did not validate the cell migration hypothesis from the mammalian claustrum or avian mesopallium.
When examining the images in Puelles et al. (2016) and other similar past studies, we noted that the patterns of the NR4A2 expression appeared quite varied and did not fill entire telencephalic subdivisions, unlike most constitutively expressed genes .
Instead, only parts of the mesopallium and hyperpallium were labeled, particularly in late developmental stages and in adults. Puelles (2017) noted that some patterns disappeared in adults. To us, the patterns appeared reminiscent of immediate early genes (IEGs), which are activated in specific cell types of brain circuits dependent on the behavior performed or sensory stimulus processed (Feenders et al., 2008;Jarvis et al., 2013;Jarvis & Nottebohm, 1997).
A literature analysis reveals that NR4A2 does undergo activitydependent expression in certain brain cell types. Most strikingly, robust NR4A2 expression induction was observed in the adult rat claustrum, deep cortical layers, in some superficial layers, and the hippocampus from 1 to 8 hr following a single subcutaneous injection of kainic acid, which induces seizure activity (Crispino, Tocco, Feldman, Herschman, & Baudry, 1998, particularly their figure 7). We noted that this induced expression pattern recapitulates much of the mouse NR4A2 expression seen in Puelles (2014) and Puelles et al. (2016). NR4A2 is part of an orphan nuclear receptor family with noted involvement in NMDAR activity-mediated and CREBdependent survival of granule cells in the rat cerebellum (Barneda-Zahonero et al., 2012;Volakakis et al., 2010). In cultured mouse hippocampal neurons, NR4A2 expression is blocked by voltage-dependent calcium channel inhibition (Tokuoka et al., 2014), indicative of activity-dependent expression. NR4A2 has a delayed expression response, as determined by qPCR and microarray assays from rat neurons, compared to its NR4A3 paralog (Saha et al., 2011). In the zebra finch, a songbird species, the NR4A1 and NR4A3 paralogs show a rapid increase in expression in several song nuclei following singing activity . Thus, the prior studies on comparative neurobiology (e.g., Puelles, 2014;Puelles et al., 2016) did not take into consideration that the NR4A2 patterns could be activity-or behavioral context-dependent. This prompted us to look into the NR4A2 brain gene expression further, using approaches we developed and used to study activity-dependent gene expression in the avian brain (Feenders et al., 2008;Jarvis & Nottebohm, 1997;Mello & Jarvis, 2008;Whitney et al., 2014).
We found both a basal and stimulus-/behavior-driven pattern of NR4A2 expression in the avian brain, confounding prior hypotheses on avian brain organization and homologies with mammals. Importantly, our findings contradict the interpretations presented in Puelles et al., 2016, do not contradict the renaming of HD as MD nor other aspects of the hypothesis that the dorsal and ventral pallial populations in the avian brain, are similar, and therefore, differing from the tetrapartite hypothesis.

| Behavioral context and sample collection
Animals were cared for in accordance with the standards set by the American Association of Laboratory Animal Care and Rockefeller University's Animal Use and Care Committee. All animals used for data analysis were collected at Rockefeller University. A total of 12 adult zebra finch males (>90 days old) were used, four for each group described below.
Following protocols we developed to measure expression of activitydependent genes in the avian brain (Feenders et al., 2008;Jarvis et al., 2013;Whitney et al., 2014), animals were placed individually in sound attenuation chambers overnight (at least 12 hr) to reduce stimulus-and behaviorally regulated gene expression to baseline levels, and then treated under the following three conditions: • Silent in darkness: Animals taken prior to the lights turning on in the morning.
• Silent in light: Animals were taken after 1.5-2.5 hr of the lights turning on in the morning, moving around, feeding, and drinking, but not singing.
• Singing: Animals were monitored and those that produced at least 25 undirected song bouts (continuous $4-20s periods of songs separated by <500 ms) per 30 min, within 1-1.5 hr after the lights turned on, were taken for the study.
After each condition was complete, animals were quickly euthanized (<1 min) by rapid decapitation, and whole brains were excised, cut mid-sagittally; separated hemispheres were embedded in block molds containing Tissue-Tek (Fisher HealthCare, Houston, TX) and quickly frozen in a slurry of dry ice powder and 100% ethanol. The amount of time between removing the bird from the sound attenuation chamber and freezing the brain tissue was under five min, so as to not measure induced gene expression due to the stress of euthanasia. Sections were cut on a CM1950 cryostat (Leica Biosystems, Buffalo Grove, IL) at 12 μm thickness in sagittal or coronal planes, and mounted onto Superfrost Plus slides (Fisher Scientific, Pittsburgh, PA).

| Single label in situ hybridization
Plasmids containing RNA polymerase promoters and cDNA sequences for NR4A2 or other genes of interest were used to amplify the cDNA inserts by PCR. We also used an alternative method, where we isolated the cDNA insert from the plasmid with PvuII-HF (New England Biolabs, Cat. The generated RNA probes were purified by ethanol precipitation, resuspended in 90% formamide, and stored at À80 C until further use.

| Double-label in situ hybridization
To determine overlap of multiple genes in the same cells, we per-

| Gene expression quantification
Images of the brain sections were taken at Â4 magnification on an Olympus BX61 upright microscope (colorimetric single-labeling) or images of the brain sections were desaturated and inverted using photoshop commands, and the signal intensity in brain regions of interest was normalized across samples by dividing by background intensity levels of a control brain region (i.e., striatum) qualitatively observed to be free of NR4A2 mRNA signal. Quantification was not completely blinded, as the NR4A2 induction was plainly visible across conditions.
The marquee tool was used to select a portion of the brain region of interest, and labeled cells were then automatically selected within the region using the Color Range selection tool to select Highlights.
The Color Range parameter (0% fuzziness) was strictly set to the lowest nonselecting Range value in control brain regions (i.e., striatum).
The number of selected cells were recorded using the Record Measurements tool. These counts were divided by the area of the selected brain region to obtain the number of labeled cells per mm 2 . Significance between groups was measured using ANOVA, followed by Tukey's HSD test for post-hoc analysis.

| NR4A2 brain expression pattern varies across behavioral conditions
In the dark housed zebra finches, there was high basal NR4A2 expression in the arcopallium (A), and in the dorsal nucleus of the hyperpallium (DNH; as seen in sagittal sections; Figure 1a). DNH is a brain region involved in night vision and magnetic field sensing in dim light conditions (Mouritsen, Feenders, Liedvogel, Wada, & Jarvis, 2005;Zapka, Heyers, Liedvogel, Jarvis, & Mouritsen, 2010).
There was also consistently high expression in a layer of cells directly above the ventricle and posterior to DNH (Figures 1a and 2a), which has been considered either as part of the hippocampus ( showed strong labeling (Figure 1a). In all remaining telencephalic regions, NR4A2 expression was very low or undetectable.
In light-exposed animals, in addition to the expression pattern seen in dark housed animals, we noted induced NR4A2 expression in more isolated cells of the visual hyperpallium ( Figure 1b) and anterior somatosensory hyperpallium (Figure 2b,c), as well as adjacent parts of the MD. We noted expression remained within Cluster N (consisting of adjacent parts of H and MD), consistent with the long decay period of NR4A2 observed in mammals (Crispino et al., 1998;Saha et al., 2011).
The induced expression in the somatosensory anterior hyperpallium and adjacent anterior mesopallium was statistically confirmed in quantitative analyses of the number of labeled cells/mm 2 compared to dark-housed animals (Figure 4a,b). Induction was found in both the MV and MD (previously named HD, Reiner, Perkel, Mello, & Jarvis, 2004) regions of the mesopallium. There was no quantitative difference seen in the intermediate arcopallium (Figure 4c). Even with these induced levels, the density of NR4A2-positive cells in the arcopallium were still higher than the induced expression in the hyperpallium and mesopallium.
In the singing animals, in addition to the patterns seen in the light housed animals (Figures 2b-g and 4a-c), induced NR4A2 expression was most notably seen in the HVC song nucleus (Figures 1c and 5a,b). Quantitative analysis revealed a 25-fold increase in mean HVC expression over dark-housed animals (Figure 5c). Singing-induced expression was also observed in the RA song nucleus (Figure 5d, e). This difference was best revealed in shorter incubation times of the chromogenic reactions, as longer exposure times needed for other pallial regions saturated the arcopallium and obscured differential expression signals in RA. In an animal that sang the most (Silver189; $104 song bouts), we also saw increased NR4A2 expression in the NIf song nucleus (Figures 1c and 5f) but not in the LMAN song nucleus, even though both are located in the nidopallium.
We noted overall low expression throughout other parts of the nidopallium in all animals, but even when present at low levels, it appeared to be adjacent to a region of higher expression in MV (Figure 2h), reminiscent of the columnar activation for other IEGs reported in Jarvis et al. (2013). We did not observe NR4A2 expression in Area X (Figure 1c) or any other striatal region across all behavioral cohorts, indicating that NR4A2 is not expressed in the avian striatum, similar to previous observations in mammals (Crispino et al., 1998;Puelles, 2014).
In both light-exposed and singing animals, we noted large variabil- In some singing animals, we noted sparse, low-level induction of NR4A2 in the caudal pallidum (CP; Figure 6a), which is directly ventral to the auditory part of the caudal striatum (CSt) that shows activitydependent gene expression responses when birds hear song (Feenders et al., 2008;Jarvis et al., 2013). We hypothesized that this induced expression in the pallidum may be due in part to the birds hearing themselves sing. Interestingly, in sagittal sections from a behaviorally undocumented mouse from the Allen Brain Atlas labeled with NR4A2, we observed very weak expression in the homologous globus pallidus ( Figure 6b). This mouse also had high expression in the claustrum and parts of layer 6b of the cortex, confirming these regions in mammals have increased NR4A2 expression (see Allen Mouse Brain Atlas [RRID: SCR_002978], NR4A2 experiment 733). Overall, these findings demonstrate that NR4A2 brain expression in birds can be activity induced, as in mammals. Such induced expression showed selective patterns, consistent with the specific functions of the brain region or circuit subset involved.

| Double labeling clarifies brain subdivision boundaries, cell types, and subcircuits activated
With limited basal expression patterns and specific activity-dependent NR4A2 expression patterns, it can be difficult to determine the full extent of brain subdivision boundaries when examining it by in situ expression alone. To more concretely verify the locations of NR4A2 cells in the hyperpallium and mesopallium regions, we performed double labeled in situ hybridization with FOXP1, a strong mesopallium marker that distinguishes MV and MD from the hyperpallium and the intercalated hyperpallium (IH) in between them ( Figure S1; Jarvis F I G U R E 1 Basal and activity-dependent induction of NR4A2 expression in the zebra finch brain. (a) Medial and lateral brain sagittal sections of NR4A2 in male zebra finches that were in the dark. (b) Sections from an animal exposed to lights for 2 hr, after an overnight in the dark. (c) Sections from an animal exposed to lights and singing at least 50 song bouts within 60 min prior to sacrifice, after overnight in the dark.
(d) Illustrations of brain subdivisions based on adjacent Nissl stained and FOXP1 labeled ( Figure S1) sections and equivalent sections of a digital brain atlas (Karten et al., 2013). Note the differences in mesopallium and hyperpallium expression across behavioral conditions. Asterisks denote regions of significant NR4A2 induction across groups, and arrows denote the somatosensory region of IH. Images are tiled at Â4 magnification, scale bar = 1 mm. Dorsal is up, posterior is left. Abbreviations and corresponding names are shown in abbreviation, Table 1 F I G U R E 2 Higher magnification of NR4A2 brain expression across conditions. is what Puelles et al. (2016) claimed to be HD, which is not our revised MD . We also noted that mesopallial cells with NR4A2 induction co-expressed FOXP1 (Figure 7a,b). The pattern of NR4A2 seen in the anterior mesopallium and hyperpallium of the active animals is reminiscent of induced IEG expression in the posterior visual and anterior somatosensory parts of these brain regions when animals are very active (Feenders et al., 2008). In coronal sections, the strongly labeled NR4A2 cells in the CDL were not within the FOXP1-bounded mesopallium, thus confirming their location more dorsally in the hyperpallium (Figure 7c). The LMI border region of higher NR4A2 expression seen in the chromogenic images (Figure 3c) was also observed within the FOXP1 boundaries (see blue arrowhead in Figure 7c 2 ) with NR4A2 expression filling portions of both MV and MD.
To verify the NR4A2-positive cells in the arcopallium, we performed double labeling with the ER81 transcription factor, a marker of avian arcopallial neurons and mammalian layer 5 projection neurons and pallial amygdala (Crispino et al., 1998;Jarvis et al., 2013). Using coronal sections, we found that both genes co-expressed in many cells of the arcopallium (Figure 7d). One exception was in the anterior arcopallium (AA) nucleus, where NR4A2 was highly expressed and  Table 1 For the singing animals, we performed double-labeling of NR4A2 with a well-studied IEG, EGR1 (Mello & Jarvis, 2008). Compared to nonsinging controls, we found double labeled NR4A2 + EGR1 cells throughout HVC (Figure 8a-c) and RA (not shown), whereas only EGR1 was expressed in the LMAN and Area X song nuclei (not shown). There was also activity-induced NR4A2 + EGR1 expression in the HVC shelf (Figure 8b,c), a nidopallial auditory area. In summary, the double labeling experiments support the hypothesis that a subset of the circuits and brain subdivisions for a particular behavior or stimulus have activity-induced NR4A2 expression.

| DISCUSSION
Our findings demonstrate that NR4A2 is an activity-dependent gene regulated by behavioral and sensory stimuli in the avian brain, as in (a-c) Quantification of NR4A2positive cells/mm 2 of dark-housed (n = 3), light-exposed silent (n = 3), and singing (n = 3) animals in the (a) anterior hyperpallium, (b) anterior mesopallium and (c) arcopallium. Shown are bar graphs, where dots represent values of individual birds, and error bars represent SEM. *denotes significance using a onefactor (behavioral context) ANOVA and Tukey's HSD post-hoc test, p < .05 mammals (Crispino et al., 1998;Saha et al., 2011;Tokuoka et al., 2014). High basal expression in the telencephalon is restricted to the arcopallium and several hyperpallial regions, and high activitydependent expression is enriched in the hyperpallium and mesopallium, and less so in the nidopallium outside of the song nucleus HVC (Figure 9a; Table 2, column 2). These clusters of induced expression cross subdivision boundaries in regions that make up circuits for specific behaviors or processing of stimuli (Horita et al., 2012;Horita, Wada, Rivas, Hara, & Jarvis, 2010;Whitney et al., 2014). While we did not investigate the time course of expression, it is possible that robust activity-dependent induction through longer durations of behavior  or systemic kainic acid experimental stimulation may further extend the expression profile of NR4A2 into areas not seen. However, such kainic acid NR4A2 induction in the rat has been limited to the claustrum, deep and superficial cortical layers, and hippocampus (Crispino et al., 1998), and this pattern was stable between 1 and 4 hr after induction, consistent with the limited regions of induction we see here in the zebra finch. It is imperative to correctly understand the functional brain organization patterns of this gene, as mutations of human NR4A2 have been linked to schizophrenia (Buervenich et al., 2000), Parkinson's Disease (Liu et al., 2017), and neuroprotection for Alzheimer's Disease (Moon et al., 2019).
There are two current competing hypotheses on avian pallial organization. The first is the discontinuum hypothesis (named as such in our companion study, Gedman et al., 2021), where the dorsal pallium regions above the vestige of the lateral ventricle (collectively called the hyperpallium) are considered distinct from the pallial regions below the ventricle (Table 2,  However, the Jarvis and Chen studies did not state that IH belonged to the mesopallium, but instead that it is a separate cell population from the hyperpallium and mesopallium, with its own molecular profile that receives heavy thalamic input similarly to intercalated nidopallium (IN; Field L2, entopallium, and basorostralis). The sharply  Table 1 further. In songbirds, there is clear evidence that DNH is a visual brain region associated with magnetic field sensation (Mouritsen et al., 2005;Zapka et al., 2009). We further advise against naming this structure ICo to avoid confusion with the long-standing abbreviation of a well-characterized avian midbrain region, the intercollicular complex (ICo; Wild, Li, & Eagleton, 1997;Kingsbury, Kelly, Schrock, & Goodson, 2011).
The NR4A2 activity-dependent induction profile we observed here is more consistent with the continuum hypothesis of cell relationships, as the patterns of expression are comparable in MD and MV on either side of the vestigial ventricle (LMI lamina). One inconsistency with this hypothesis is the weaker induction of NR4A2 in the nidopallium compared to the hyperpallium. However, the NR4A2 label in the hyperpallium appears in sparsely labeled cells, which has been seen for only a few other genes (e.g., SATB2) that differentially label the hyperpallium relative to the nidopallium, and may point to a unique cell type in the hyperpallium. This is discussed further in our companion study (Gedman et al., 2021). There, we also provide further support for the Jarvis et al. (2013) and Chen et al. (2013)  ing the context of another two competing sets of hypotheses on brain homology between birds and mammals Jarvis et al., 2013;Reiner et al., 2004). The first of these is the nuclear-to- However, the basal levels of NR4A2 in the claustrum and activityinduction patterns in the adjacent deep cortical layers (Crispino et al., 1998) (Table 2; Puelles et al., 2000). There was also no interpretation offered for the high basal NR4A2 levels in the avian arcopallium, but not in the proposed mammalian homolog, which they and others designate as the pallial amygdala.
For alternative explanations, one could argue for species differences between chicken (Puelles et al., 2016) and zebra finch (this study), though Wang et al. (2011) analyzed NR4A2 (Nurr1) expression in adult chickens and found labeling mostly restricted to the F I G U R E 9 Brain subdivision boundaries of basal and activity-dependent induction of NR4A2 expression in the avian and mammalian brains.
(a) Model of the extent of brain regions with high (dark green) and low (light green) activity-induced NR4A2 expression the avian brain, using the songbird as an example. (b) Model of NR4A2 expression in the mammalian brain, using the mouse as an example. Expression bounds are based on this study, Crispino et al., 1998, Puelles, 2014, Puelles et al., 2016. Further details are in Table 2. Abbreviations and corresponding names are shown in abbreviation, Table 1 T A B L E 2 Competing models of brain organization and homology compared to NR4A2 expression hyperpallium, without labeling in the mesopallium, reminiscent of patterns we see here in some zebra finches. Some patterns may also be recapitulated across multiple individuals at the same developmental stage, which we believe would more likely demonstrate a stereotypical activity pattern of developing neural circuits (Ant on-Bolaños et al., 2019) than distinct brain subdivision boundaries.
Given that the patterns are not consistent with the homology arguments in the Puelles et al. studies or a tetrapartite avian brain organization, we wonder whether there is support for alternative hypotheses. One could interpret the high NR4A2 expression patterns in the mammalian deep cortical layers as consistent with homology to the avian arcopallium ( Table 2). The weaker induction in the superficial cortical layers of mammals (Crispino et al., 1998) would be consistent with homology to the avian mesopallium and hyperpallium. But these anatomical delineations are based on only one activity-dependent gene. Two studies, one using in situ hybridization expression profiles of seven critical transcription factors for cortex development (Suzuki & Hirata, 2014) and the other using micro-array expression profiles of over 7000 orthologous genes , concluded that mammal cortex layers 2 and/or 3 were most molecularly similar to the avian mesopallium or nidopallium, respectively.
There are other claims made in Puelles et al., 2016 and subsequent studies (Puelles, 2017;Watson & Puelles, 2017) using NR4A2, which we argue are confounded without proper consideration of the animal's activity state. The importance of controlling animal behavior state and awareness of a gene's activity-regulated expression have been discussed and demonstrated in past studies, including for birds Mello & Jarvis, 2008). Interpretations can change dramatically when taking brain activity states into consideration. This is still necessary to consider in embryos despite difficulties in controlling behavior in ovo. With our demonstration of differences in interpretation between studies using only this one gene, we hope that future studies will more seriously take behavioral and stimulus context into consideration.

ACKNOWLEDGMENTS
We thank Irene Ballagh, Jun Takato, Rebecca Stephen, and Katherine Farber for assistance on the optimization and execution of the nonradioactive in situ hybridization protocols. Confocal images were taken using resources from The Rockefeller University Bio-Imaging Resource Center. This project was funded by the Howard Hughes Medical Institute and Rockefeller University start-up funds to Erich D. Jarvis.

PEER REVIEW
The peer review history for this article is available at https://publons. com/publon/10.1002/cne.25157.

DATA AVAILABILITY STATEMENT
The data supporting the findings of this paper are primarily presented within the scope of this publication. Additional images and materials are available upon request to the corresponding author (EDJ).