Egr1 Is Necessary for Forebrain Dopaminergic Signaling during Social Behavior

Egr1 influences overall growth but not development

To identify a role for the egr1 gene in social behavior, we first examined the expression pattern of egr1 in developing zebrafish larvae by RNAscope ISH at 6 dpf. This developmental period, during which synaptic circuitry remains dynamic (Goode et al., 2020), occurs one week before the appearance of robust social behavior (Dreosti et al., 2015; Larsch and Baier, 2018; Stednitz and Washbourne, 2020). egr1, in addition to being expressed in the pharyngeal pouch (PP; Fig. 1A), is predominantly expressed in a few clusters in the forebrain and midbrain. Given previous evidence for telencephalic contribution to social behavior (Stednitz et al., 2018), we noticed the prominent egr1 expression in the telencephalon, including the BF, in cells adjoining the anterior commissure (AC; Fig. 1B). We conclude that egr1 is expressed in a region that contributes to social behavior at a time when neuronal circuits are being established.

We obtained a mutant from the Sanger ZMP with a point mutation in the egr1 gene (sa64). The C > A mutation generates a premature termination codon in exon 2, likely resulting in a truncated 137-aa protein. As this mutation is located upstream of the functional zinc finger DNA binding motifs (Fig. 1C), we predict that it results in loss of protein function. To eliminate possible additional background mutations, we used egr1^sa64 hets that had been outcrossed to wt zebrafish for seven generations.

Larvae derived from an incross of egr1^sa64 hets develop normally (Fig. 1D) and homozygous mutants were present at an approximate Mendelian ratio at 15 dpf (27.7 ± 13.6% SD; N = 4 clutches, n = 201 animals), a developmental time at which social behavior is already robust (Stednitz and Washbourne, 2020). However, we noticed a large proportion of larvae that were much smaller (< the mean minus 1.5 × the SD) than others, which we will call “stunted.” Mutants were significantly smaller than wt and hets on average (p = 0.002, p = 0.006), although stunted individuals were found across all genotypes and reached other developmental milestones with their siblings (Fig. 1D, bottom). For example, stunted larvae inflated their swim bladders by the same day (5 dpf) as all other animals, including homozygous egr1^sa64 mutants. Although the majority of homozygous egr1 mutants were stunted (78 ± 12.9% SE across 8 clutches; Fig. 1E), almost half of the stunted larvae occurred in het and wt larvae (Fig. 1F). This phenotype is similar to the stunted growth reported in the mouse Egr1 gene knock-out mutant (Topilko et al., 1998). Analysis of egr1 using morpholine-modified oligonucleotides (MO) in zebrafish also resulted in reduced growth in morphants (Zhang et al., 2013). Therefore, despite the incomplete penetrance in our system, we conclude that egr1 contributes to size of the animal. Because of the relatively high occurrence of the stunted phenotype in hets, we conclude that this phenotype may be because of a gene interaction with an, as yet, undetermined other gene. We also conclude that egr1^sa64 is likely to be a null allele, although we cannot exclude the possibility that it is a hypomorph. Importantly, we considered size as a factor for all additional experiments.

Egr1 is necessary for social behavior

To examine social behavior in egr1^sa64 mutants of regular-sized and stunted animals, we used a split dyad assay with size-matched animals (Fig. 2A; Stednitz et al., 2018). Social behavior becomes robust at 14 dpf and is visually driven, with animals engaging in stereotypical orienting and approach behavior when they can see an animal of similar size (Dreosti et al., 2015; Stednitz and Washbourne, 2020). egr1 mutants spent less time orienting at angles to the divider between 45° and 90° (p = 0.003; Fig. 2B), a strong indicator of social engagement (Stednitz et al., 2018), and less time close to the divider (Fig. 2C). Mutants were not less active during the assay period (p = 0.131; Fig. 2D). We conclude that Egr1 is involved in modulating social behavior.

Consistent with our previous findings on active reciprocation during social interactions in adult and larval zebrafish, wt and heterozygous fish paired with mutants also exhibited reduced social orienting (p = 0.003; Fig. 2E) and place preference (p < 0.001; Fig. 2F). Using multiple linear regression, we found that both the genotype of the target fish and the stimulus fish significantly influences orienting behavior (p = 0.021 and p = 0.040, respectively). Therefore, the deficit in social behavior of egr1 mutants is detected by wt and heterozygous siblings. However, we also conclude that this social assay provides a variable social stimulus for test larvae that might lead to spurious results, especially with the added size effect.

To explore social engagement without stimulus variability, we tested whether a social deficit could also be detected using a virtual social assay that is based on detection of biological motion (Larsch and Baier, 2018). This assay reduces variability, as the stimulus is identical for each larva. Dots projected below dishes of solitary animals are followed preferentially when the dots move in bouts with kinetics similar to age matched animals (Fig. 3A; Larsch and Baier, 2018). Animal responses are tuned to dot size, with maximal social attraction at dot sizes approximating body size at 4.0 mm for 14 dpf larvae (Fig. 3B; Larsch and Baier, 2018). Larger dots (16 mm) elicit an aversive response, whereas small dots (0.5 mm) are ignored and are equivalent to no dot (0.0; Fig. 3B). Homozygous egr1^sa64 mutant larvae are less attracted to visual stimuli 2.0, 4.0, and 8.0 mm in diameter than wt siblings (p = 0.0086, 0.0034, 0.047, HSD; Fig. 3B). This reduction in social behavior is also evident at 16 and 18 dpf (p < 0.0001, p = 0.009, HSD), suggesting the deficit is not a short developmental delay that is recuperated over 4 d of development (Fig. 3C). egr1 genotype does not affect animal swimming speed or thigmotaxis (p = 0.69, 0.76, HSD; Fig. 3D,E). Furthermore, predator avoidance responses to a looming stimulus are not significantly different between all genotypes at 14 dpf (p > 0.25; HSD; Fig. 3F), suggesting that vision is not accountable for the reduction in SI. We conclude that Egr1 is necessary for following biological motion in a social context, but not for visual and motor system functions.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

egr1 mutants are less social with virtual stimuli than wt larvae. A, Biological motion-based social behavior was assayed by projecting dots of different sizes to dishes from below, while recording with an IR camera from above. B, Violin plots revealed significant differences for egr1 mutants in SI for dot sizes of 2.0, 4.0, and 8.0 mm at 14 dpf. N = 3 experiments, n = 75 animals *p < 0.05, HSD, wt (+/+, white), het (+/−, light gray), mut (−/−, dark gray). C, egr1 mutants exhibited social deficits at 16 and 18 dpf, shown for optimal dot size 4.0 mm (dot size with highest SI response). N = 3 experiments, n = 129 animals, and N = 3 experiments, n = 97 animals, *p < 0.05, HSD. Plots of average speed in pixels per second (D) and thigmotaxis index (E) during biological-motion based assays at 4.0 mm dot size revealed no general behavioral deficits for egr1 mutants. N = 3 experiments, n = 77 animals. F, Mutant predator avoidance responses were not significantly different from het and wt siblings across various stimulus sizes at 14 dpf. N = 3 experiments, n = 126 animals. G, Correlation plot of size (in pixels) and SI for 4 mm dot size at 14 dpf. N = 3 experiments, n = 75 animals. H, The differences in SI for 4.0 mm dot size in 14 and 16 dpf larvae (normalized to average 14 dpf wt SI) between mutants and wt/hets (+, white) are significant in both regular-sized and stunted larvae. n = 201 animals, *p < 0.05, HSD. Error bars are SEM. Animal numbers are presented on bars.

We considered that the growth deficit (Fig. 1D–F) might account for the reduction in SI. We detected a weak, yet significant, correlation between animal size and SI (R = 0.37, R² = 0.14, p = 0.0011; Fig. 3G). This effect was consistent across individual genotypes and at all ages examined. We further explored the influence of size by comparing the SI of animals in groups of similar size. Consistent with previous observations (Larsch and Baier, 2018), larger wt and heterozygote animals have a 67% higher SI than wt stunted animals (p = 0.186, HSD, n = 9 stunted animals; Fig. 3H, +). However, mutants exhibited reduced social behavior both for regular-sized and stunted individuals (p = 0.0001, HSD; Fig. 3H, −/−). We conclude that genotype, rather than size, is the primary factor influencing social behavior deficits in egr1 mutants.

Egr1 is necessary for TH2 expression

The Egr1 transcription factor regulates expression of proteins that play a role in synaptic transmission and NT synthesis (Duclot and Kabbaj, 2017). We used Western blotting of protein extracts from 15 dpf larval heads to explore whether Egr1 promotes social behavior by influencing abundance of such proteins in the brain. Levels of MAGUK proteins (panMAGUK), such as PSD-95, SAP102 and SAP97, and Synapsin proteins 1 and 2 (Synapsin 1/2) were not significantly influenced by egr1 loss of function (p > 0.8, HSD; Fig. 4A–C). In contrast, an antibody that recognizes two forms of TH, which based on predicted molecular weights likely correspond to TH1 and TH2, revealed that levels of TH2 but not TH1 are compromised by mutation in egr1 (TH1: p = 0.9, TH2: p = 0.0485, HSD; Fig. 4A,A’,D,E).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Egr1 is necessary for TH2 expression. A, Western blotting of larval head homogenates at 14 dpf reveals a reduction in expression of TH2 in egr1 mutants. A’, Two color labeling of a Western blotting to highlight in-lane normalization to β-actin and technical replicates. B–E, Western blot intensities in A, A’ were first normalized to in-lane β-actin, and then to wt average intensity. N = 4 experiments, n = 2 technical replicates, *p < 0.05, HSD. F, Immunolabeling of synapses with antibodies to Synapsin 1 and 2 (Syn1/2, magenta) and Gephyrin (Geph, green) in the anterior commissure (AC), lateral view. Insets are magnifications of the cyan box. Scale bar: 5 μm. G, Quantification of synaptic puncta revealed no changes in synapse density in the medial and lateral AC. n ≥ 5 brains per genotype. wt (+/+, white), het (+/−, light gray), mut (−/−, dark gray). Error bars are SEM.

Given the bias of Egr1 toward regulating synaptic genes (Duclot and Kabbaj, 2017), we assessed whether Egr1 influences forebrain synapse abundance. We quantified synapses by immunolabeling sections of 15 dpf larval brains with antibodies to Synapsin 1/2 and Gephyrin in egr1^sa64 mutants (Fig. 4F). Synapsins are enriched at over 95% of vertebrate synapses (Micheva et al., 2010), whereas Gephyrin is a specific marker of inhibitory synapses (Fritschy et al., 2008). Within the BF, a region with high expression of egr1 (Fig. 1A,B) and a region implicated in controlling social behavior (Stednitz et al., 2018), we focused on the neuropil of the AC. Quantification of the number of puncta, an approximation of synapses, in the medial and lateral AC neuropil showed no difference in total (Synapsin 1/2, p = 0.71, HSD) or inhibitory (Gephyrin, p = 0.99, HSD) puncta density in egr1 mutants compared with wt (Fig. 4G). The lack of effect on synapse number is consistent with the Western blotting results above, which demonstrated that levels of the widespread synapse markers Synapsin 1/2 and PSD-95 are unchanged in egr1 mutants. However, our results do not preclude the possibility that specific synapse populations might be regulated by Egr1. We conclude that Egr1 regulates the abundance of TH2 protein, but does not regulate presynaptic or postsynaptic protein abundance of Synapsin 1/2 or MAGUK proteins, nor general synapse density.

A synaptic connection between TH-positive neurons and cholinergic BF neurons (cBFNs) defined by the enhancer trap line Et ^y321 [Et(rex2-scp1:gal4ff)y321] was recently identified (Goode et al., 2020). Chemogenetic ablation of cBFNs disrupts social attraction and orienting behavior (Stednitz et al., 2018). Images from the most basal region of the forebrain (Fig. 5E), including the anterior part of the presumptive PPa, reveal that the TH-positive neurons project to the AC neuropil (Fig. 5A–A’’, white arrows), which is also innervated by cBFN arbors (Goode et al., 2020). TH-positive neurons in the PPa are highly likely to be dopaminergic (Yamamoto et al., 2011), so we will refer to these cells as dPPaNs. Cells in this region have been documented to express both th1 and th2 genes in larvae and adults (Semenova et al., 2014).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Egr1 is necessary for TH-positive neuron number in BF. A–A’’, Dorsal view of maximum projection of the BF region (magenta box in E) of Et ^y321 larval zebrafish at 14 dpf labeled with antibodies to GFP (green) and TH (magenta). Anterior commissure (AC); anterior is to the top. B–D, Maximum projections of TH labeling in the BF region posterior to the AC for egr1 wt (B), hets (C) and mutants (D). Scale bars: 30 μm. E, Diagrammatic side-view of the larval brain highlighting the regions (magenta and cyan) analyzed as a maximum projection Z-stacks in A, B–D, respectively. F, Quantification of the number of TH-positive neurons in the cyan region in E revealed a decrease in egr1 mutants at 14 dpf. N = 2 experiments, n = 18 brains, p < 0.05 (HSD). G, The same quantification as in F at 7 dpf demonstrates a consistent decrease in TH-positive neurons during development. N = 2, n = 24, p < 0.05 (HSD). wt (+/+, white), het (+/−, light gray), mut (−/−, dark gray). Error bars are SEM.

We explored whether TH expression in dPPaNs requires Egr1 (Fig. 5B–D). The number of TH-positive dPPaNs was significantly reduced in homozygous egr1^sa64 mutants relative to wt siblings at 14 dpf (p = 0.049, HSD; Fig. 5F). The ∼28% reduction of TH-positive dPPaNs in mutants was evident as early as 7 dpf (Fig. 5G), a time at which stunted individuals were not detected. Interestingly, we found an intermediate, but not significant, phenotype in TH cell number in heterozygous egr1^sa64 mutants as compared with wts at both time points (p = 0.362, p = 0.099, HSD), analogous to results from the Western blot experiments (Fig. 4E). This may indicate a haplo-insfficiency at the molecular level, although this is not detected at the behavioral level. Together, our results suggest that the discrepancy in TH expressing cells is established early in development, consistent with the expression of egr1 in the BF at this developmental stage (Fig. 1B). Given the expression of both th1 and th2 in dPPaNs (Semenova et al., 2014), we conclude that our result is because of a decrease in the actual number of dPPaNs, and not because of a detection limit as a consequence of a th2 expression decrease. Our experiments suggest that catecholaminergic, presumably dopaminergic, signaling in this part of the larval brain is compromised by loss of Egr1, although it is possible that feedback mechanisms might compensate for the NT deficit.

BF TH neurons are necessary for social behavior

To gain genetic access to the TH-positive neuron population and test a possible role for these neurons in social behavior, we examined driver lines that express in the BF. The Tg(galanin:GFP) line expresses in neurons in the PPa, and the neuropeptide galanin has been implicated in various mammalian social behaviors (Wu et al., 2014). Labeling of galanin:GFP transgenic animals with TH antibody revealed that galanin-expressing and TH-positive neurons are distinct populations (Fig. 6A). The enhancer trap line Et ^y405 [Et(scp1:gal4ff)y405], in which the construct integrated in the robo2 gene, also drives expression in the PPa. In contrast to the galanin driver, ∼30% of neurons labeled by the Et ^y405 driver overlapped with TH-positive dPPaNs (29.2 ± 5.1%, N = 3 brains; Fig. 6B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

TH-positive neurons in PPa are necessary for social behavior. A–A’’, Maximum projection of the PPa of 14 dpf Tg(galanin:GFP) larvae immunolabeled for GFP (green) and TH (magenta) revealed two distinct populations. Dorsal view of cyan box in Figure 5E, anterior to the top. B–B’’, Maximum projection of PPa of 14 dpf Et ^y405 larvae immunolabeled for GFP (green) and TH (magenta) revealed transgene expression in ∼30% of TH-positive neurons (yellow asterisks). Scale bars: 20 μm. C, Violin plots of SI for Et ^y405 larvae with nitroreductase-ablated neurons at 16 dpf (magenta) revealed significant differences compared with control larvae only expressing GFP (green). Both groups were treated with nifurpirinol. N = 2 experiments, n = 60 animals, *p < 0.001, t test comparisons to GFP control. Plots of average speed in pixels per second (D) and thigmotaxis index (E) during biological-motion based assays at 4.0 mm dot size revealed no general behavioral deficits for larvae with Et ^y405 neurons ablated. N = 2 experiments, n = 60 animals, p > 0.5. Error bars are SEM.

We hypothesized that chemogenetic ablation of neurons marked by Et ^y405 would reduce dPPaNs similar to egr1 loss of function, allowing us to test whether a reduction of dPPaNs might be the cause of reduced social behavior in egr1 mutants. We treated larvae generated by a cross between the Et ^y405 driver and a nitroreductase expression line with the antibiotic nifurpirinol. The enzymatic activity of nitroreductase on nifurpirinol generates a toxic by-product that kills cells in a cell-autonomous fashion. Indeed, treatment of animals with nifurpirinol resulted in a 31% reduction in dPPaNs when the nitroreductase transgene was expressed (GFP control: 82 neurons ± 2.5, nitroreductase: 56.2 ± 5, n = 11 brains, p < 0.005, t test). Biological motion-based social behavior was significantly reduced in larvae with ablated dPPaNs compared with control animals at dot sizes 2.0, 4.0, 8.0, and 16.0 (p < 0.001, t test; Fig. 6C). No deficits were detected for swimming speed or thigmotaxis (p > 0.5, t test; Fig. 6D,E). Chemogenetic ablation of random neuronal populations in larvae does not always cause a deficit in social behavior (Stednitz et al., 2018), suggesting that the social behavior deficit is specific to the ablated population and not because of general neurologic trauma. We cannot exclude that ablation of neurons in other regions of the brain within the Et ^y405 line might contribute to the social deficit, as the Et ^y405 enhancer trap drives sparse expression throughout the brain at 16 dpf (Fig. 7). However, our results are consistent with the conclusion that dPPaNs are necessary for robust social behavior.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Whole-brain expression of Et ^y405 at 16 dpf. Maximum Z projection of Et ^y405 (y405) brain expressing GFP, with the region of interest from Figure 6B marked with a cyan box. Anterior to the top; * nonspecific labeling of meninges. Tel, telencephalon; PPa, anterior parvocellular POA; OT, optic tectum; Th, thalamus. Scale bar: 200 μm.