Molecular signature of excessive female aggression: study of stressed mice with genetic inactivation of neuronal serotonin synthesis

The World Health Organization (WHO) categorizes violent behavior among the top 20 causes of disability worldwide and this is likely to increase by 2030 (Vakili et al. 2015). Recently, there have been increasing numbers of reports of females displaying aggression associated with neurodevelopmental and psychiatric disorders (Freitag et al. 2018) and rising female criminality (Denson et al. 2018). Although women are generally less aggressive than men, excessive aggression in women can be harmful, leading to devastating consequences for society, such as misconduct, family conflicts and delinquency, including severe physical injury and homicide (Xiang et al. 2019). Nevertheless, the neurobiology of aggression, particularly in females, remains elusive. This necessitates the need to understand the molecular mechanisms underpinning female aggression.

Decreased synthesis of brain serotonin (5-HT) has been demonstrated to play a critical role in pathological aggression in both sexes. The key enzyme of neuronal 5-HT synthesis is tryptophan hydroxylase-2 (TPH2) (McKinney et al. 2005). Low expression of the TPH2 can be caused by such common single nucleotide polymorphisms (SNPs) in the gene’s transcriptional control region (Herrmann et al. 2007) as c.‐1449C > A (rs7963803) and c.‐844G > T (rs4570625), -7180T > G, -7065C > T, -6526A > G, and -5806G > T. These SNP variants in TPH2 gene have been implicated in psychiatric disorders presenting with high anxiety and impulsive aggression (Rotondo et al. 1999; Carkaci-Salli et al. 2006; Oades et al. 2008; Plemenitaˇs et al. 2015; Laas et al. 2017), personality disorder (Perez-Rodriguez et al. 2010), negative emotionality (Lesch et al. 2012), depression (Wigner et al. 2018) and suicidality (Ottenhof et al. 2018). These clinical observations have been recapitulated in translational studies in mice with complete inactivation of Tph2 (Tph2 − / −), which resulted in excessive aggression regardless of sex (Gutknecht et al. 2012, 2015; Angoa-Pérez et al. 2012; Weidner et al. 2019). Notably, the experience of stress may increase the incidence of aggression (Forssman et al. 2014; Xu et al. 2016). Moreover, aggression can also be triggered per se by environmental adversities and stressors (Haller et al. 2014; Takahashi and Miczek 2013; Xiang et al. 2019; Strekalova et al. 2021a, b), which are also known to have sex-specific effects (Takahashi and Miczek, 2013; Gorlova et al. 2023). Finally, there is accumulating evidence for the relevance of gene x environment interaction in neuropsychiatric conditions in general (Lesch 2005; Caspi et al. 2002; Lesch et al. 2012; Lesch and Mössner 2006; Xiang et al. 2019; Lee et al. 2019) and in the regulation of pathological aggression, in particular, suggesting a role for genetic variation and stress factors. However, there are only a limited number of available experimental models for studying aggression in the context of gene x environment interaction (Gorlova et al. 2023). Moreover, there is a lack of models for female aggression, which limits our understanding of the molecular mechanisms.

We recently described a mouse model of female aggression, in which mice with partial genetic inactivation of Tph2 (heterozygous Tph2^+/− mice) were subjected to predation stress, resulting in pathological aggressive behavior (Svirin et al. 2022a) reminiscent of null mutant (Tph2^−/−) mice (Gutknecht et al. 2012, 2015). This was accompanied by abnormal expression of 5-HT receptors as well as changes in biomarkers of cellular distress, myelination and neural plasticity (Svirin et al. 2022a). However, these changes were not observed in naïve Tph2^+/− mice. Compared to Tph2^−/− mutants, which lack 5-HT synthesis in the brain's raphe nuclei, Tph2^+/− mice display only a 10–30% reduction of 5-HT levels (Gutknecht et al. 2012), whereas five days of rat-exposure stress induced aggressive behavior in both male and female mice that was comparable to the effects of complete absence of brain 5-HT (Gorlova et al. 2020; Strekalova et al. 2021a, b; Svirin et al. 2022a). Specifically, in the prefrontal cortex (PFC), stressed female mutants displayed altered gene expression of 5-HT receptors, Htr1a and Htr2a, glycogen synthase kinase-3β (Gsk-3β), c-fos, and myelination-related transcripts, including myelin basic protein (Mbp), proteolipid protein 1 (Plp1), myelin-associated glycoprotein (Mag), and myelin oligodendrocyte glycoprotein (Mog) as well as plasticity markers, including synaptophysin (Syp) and cAMP response element binding protein (Creb) (Svirin et al. 2022a).

In a similar paradigm of 5-day predation stress in male Tph2^+/− mice, mutants showed excessive aggressiveness with pathological behavioral patterns of impulsive aggression, which was accompanied by changes in the concentration and metabolism of 5-HT (Gorlova et al. 2020), and abnormal levels and turnover rates of dopamine and norepinephrine (Strekalova et al. 2021a, b). Remarkably, behavioral changes in male Tph2^+/− mice were found to be the opposite to those found in Tph2^+/+ mice exposed to the predation, i.e., the latter display a decrease in aggressive and dominant behaviors under these conditions (Gorlova et al. 2020; Strekalova et al. 2021a, b). Besides, under challenging conditions, unlike Tph2^+/+ mice, male Tph2^+/− mutants did not show stress-induced changes in the floating behavior in the forced swim test, and stressed female Tph2^+/− mice had unaltered signs of anxiety and increased novelty exploration (Svirin et al. 2022a). Thus, behavioral responses to various environmental challenges in Tph2^+/− mice of both sexes were strikingly distinct from those of wild type mice that let us to hypothesize also different transcriptomic signatures of the response to stressors of the two genotypes.

The aim of this study was to examine the potential changes of the transcriptome in the brain underlying distinct behavioral responses including excessive aggression following subchronic predation stress and food deprivation in female Tph2^+/− mice. We performed deep sequencing to investigate changes in the PFC, a region of the brain implicated in both aggression and the response to stress (Arnsten 2009; Wall et al. 2012; Takahashi et al. 2014; Achterberg et al. 2016). Previous studies showed that stressed Tph2^+/− mice displayed significant changes in the PFC including 5-HT turnover (Gorlova et al. 2020), 5-HT receptors and markers of distress, neuronal plasticity, and myelination (Svirin et al. 2022a). Furthermore, the elevated 5-HT turnover in the PFC correlated with the degree of aggressiveness in stressed male Tph2^+/− mice (Strekalova et al. 2021a, b).

In this study, we examined social/aggressive-like behavior in stressed, food-deprived female mice using previously established tests of home-cage sociability and food competition (Svirin et al. 2022a). Here, stressed Tph2^+/− female mice revealed not only increased signs of dominant and aggressive behavior, but surprisingly, also killing behavior. We investigated the molecular correlates of this behavior in the prefrontal cortex using previously developed deep sequencing and EBSeq methods, followed by a global downstream and regulon analyses using standard and custom packages/scripts. There were 26 genes that appeared to be significantly altered in opposite directions in the stressed groups of both Tph2 genotypes. Based on previous findings reported in the literature, selected genes of interest associated with excessive aggression were further explored to establish relationships between these genes and the 5-HT system.

Our findings demonstrate baseline differences in escape and anxiety-like behavior of female Tph2^+/− mice as well as excessive aggression and home-cage dominant behaviors in food-deprived female Tph2^+/− mice subjected to predation stress, thus resembling a behavioral profile reported for stress-naïve Tph2^−/− female mutants. Stressed wildtype controls and non-challenged, food-deprived Tph2^+/− mice did not show any of these behavioral changes. We identified 26 genes whose expression was changed in opposite direction between the stressed groups of both genotypes. Subsequent analysis of physiological functions of these genes revealed several molecular markers presumably mediating genetic brain 5-HT deficiency in interaction with adverse experience that resulted in pathological aggression of female mice. Several of the DE genes, including Dgkh, Arfgef3, Kcnh7, Grin2a, Tenm1 and Epha6 have been implicated in neurodevelopmental processes, neural plasticity, metabolic turnover, 5-HT system function and regulation of stress responses;. The molecular signature of the behavioral stress response of Tph2^+/− mice identifies risk genes in pathways leading to cognitive impairment and emotional dysregulation as well as related neuropsychiatric disorders.

Unexpectedly, food-deprived stressed Tph2^+/− female mice displayed killing behavior of their conspecifics that can be interpreted as a manifestation of extreme aggression. While according to classical ethology, killing behavior is rarely occurs among conspecifics (Lorenz 1966), adult mice and rats can display this form of excessive aggression under certain extreme conditions (Lane-Petter 1968), e.g. various forms of environmental stress (Jerussi and Hyde 1985; McCarthy and Saal 1986; Zafar et al. 2018), malnutrition and dietary restrictions (Hilakivi-Clarke et al. 1996), territorial limitations (Shimamoto 2018; Weber et al. 2017), exposure to toxins (Schardein et al. 1978; Marsman et al. 1995), disturbed 5-HT-ergic (Parmigiani and Palanza 1991) and dopaminergic (Carter et al. 2002) monoamine brain regulation. This form of behavior is also called ‘intraspecific predation’ and often involves cannibalism (Lane-Petter 1968). Reported here for the first time killing behavior of Tph2^+/− female mice suggests the role of partial genetic deficit in Tph combined with subchronic stress and restricted access to food in extreme female aggression.

Among the DE genes, several were previously shown to be implicated in neurodevelopmental and brain plasticity processes and associated with increased risk for emotional dysregulation. For example, DGKH encodes diacylglycerol kinase eta, which is involved in the phosphorylation of diacylglycerine (DAG). DAG participates in the activation of protein kinase C (PKC). Thus, DGKH may contribute to the downregulation of the 5-HT2A receptor and 5-HT transporter (SERT) (Anji et al. 2001). Regulation of these critical mediators of serotonergic transmission may alter levels of extracellular 5-HT and neuronal excitability. High levels of expression were observed in brain regions implicated in cognition and emotionality. While inactivation of Dgkh in mice increases levels of dopamine in the midbrain (Asami et al. 2021), variants of DGKH were associated with ADHD and bipolar disorder (Weber et al. 2011; Whalley et al. 2012). Taken together, the upregulation of Dgkh gene may contribute to the observed increased aggression and anxiety-like behavior by impacting both serotoninergic and dopaminergic transmission. Altered expression of ARFGEF3 that encodes Brefeldin A-inhibited guanine nucleotide-exchange protein 3 (BIG3) has been linked to metabolic regulation. The protein is localized in insulin- and glucagon-containing granules in pancreatic beta- and alpha-cells as well as in synaptic vesicles of neurons. Depletion of ARFGEF3 in hippocampal neurons affects lysosomal biogenesis by reducing the number of lysosomes (Liu et al. 2016). This increases levels of GABAA receptors leading to increased GABA-dependent currents and thus enhanced GABA signalling (Liu et al. 2016). Since GABA receptors have been shown to mediate fast synaptic inhibition in neural circuits (Hsu et al. 2003) implicated in the behavioral responses to stress (Paredes and Ågmo 1992), Arfgef3 upregulation may contribute to the anxiety and stress-related behavior observed in Tph2^+/− mice. KCNH7 encodes Kv11.3 that represents the voltage-gated potassium channel, subfamily H, member 7. The highest levels of expression are observed in limbic and cortical areas involved in the modulation of cognition and mood (Strauss et al. 2014). Dysfunction of potassium currents caused by KCNH7 inhibition leads to prolongation of plateau potentials of dopaminergic neurons. Variants of KCNH7 were associated with a wide range of neuropsychiatric disorders, including developmental delay, bipolar disorder and schizophrenic psychoses (Griswold et al. 2015). Altered plateau-potentials of dopaminergic neurons may change the release of mesolimbic dopamine, thus contributing to the abnormal behavior observed in Tph2^+/− mutants. GRIN2A encodes the GLuN2A subunit of N-methyl-d-aspartate (NMDA) receptors. NMDA receptors (NMDARs) are highly expressed in the hippocampus and neocortex, localize at synaptic sites and potentially interact with teneurins (Sando et al. 2019; Franchini et al. 2020). NMDARs play a critical role in modulating excitatory synaptic transmission, affecting normal neuronal development and synaptic plasticity involving long-term potentiation as well as cognition, learning and memory (Strehlow et al. 2019). Upregulation of the GRIN2A expression results in increased NMDAR-mediated currents, thus modulating excitation/inhibition (E/I) balance (van Rhijn et al. 2022). Variants in GRIN2A were associated with developmental and epileptic encephalopathy, disorders of the epilepsy aphasia spectrum, speech disorders and schizophrenic psychoses (Strehlow et al. 2019; Franchini et al. 2020; Poltavskaya et al. 2021; van Rhijn et al. 2022). Ionotropic glutamate receptors of the NMDA subtype have been implicated in cognitive flexibility. The observed upregulation of the Grin2a gene may trigger hyperactivity of the dopaminergic network, manifested by increased excitability of mutant animals (van Rhijn et al. 2022). TENM1 encodes teneurin transmembrane protein 1, which is mainly expressed during the embryonic development of neurons in hippocampus and amygdala. Protein activity promotes actin cytoskeletal remodelling and directs neuronal connectivity. In mice, it was shown that TENM1 might participate in synaptic-matching of axons of the olfactory system and inactivation of the gene in mice results in alterations associated with olfaction (Alkelai et al. 2016). Upregulation of TENM1 contributes glioblastoma cell migration and invasion of the surrounding environment (Talamillo et al. 2017). Finally, TENM1 has been implicated in X-linked intellectual disability (Bengani et al. 2021). The upregulated genes may be operational during developmental differentiation of neurons in the hippocampus and amygdala, since both brain regions are central to emotional responses, including increased impulsivity and aggression. EPHA6 encodes ephrin-type-A receptor 6, a member of the ephrin receptor family. EPHA6 is involved in processes of brain development and specifically participates in the regulation of the cytoarchitecture of cortical neurons (Das et al. 2016). EPHA6 displays high levels of expression in the hippocampus and various cortical regions. Knockout of the gene in mice results in learning impairments and memory deficits (Alkelai et al. 2016). Epha6 knockout mice display impairments in contextual fear memory (Dines and Lamprecht 2016). Disruption of the EphA/ephrin-A signaling in mice also results in a reduction of of dopaminergic neurons and reduced volume of the substantia nigra (Sieber et al. 2004). Taken together, the alteration in the Epha6 gene expression can affect cognition and memory, thus contributing to anxiety-like behavior. In addition, the fact that a number of DE genes belong to the protein and chromatin modifier class suggests that an epigenetic regulatory mechanism is involved. It might be attractive to perform mass-spectrometry on histones to identify any differences in the histone modification landscape in mutant versus wildtype mice under stress conditions. Based on mass-spectrometry results, Chromatin immunoprecipation (ChIP) may be performed on the relevant histone modifications to identify genomic loci that are being differentially regulated in the mutants under stress conditions.

mRNA expression analysis showed that 17 regulons were significantly altered in Tph2^+/− mutants exposed to predation stress, while no effects of stress were found in wildtype mice. Among altered regulons, 12 regulons were upregulated, whereas five were downregulated. A greater number of upregulated genes were reported earlier in the brain of a subgroup of C57BL6 mice with higher baseline aggression that following exposure to chronic stress displayed stress resilience by not developing the depressive-like syndrome, in contrast to low aggressive mice that were stress-susceptible (Strekalova et al. 2004, 2011). These data echo with reported phenotype of increased aggressive traits and signs of stress resilience in mice with Tph2 genetic deficiency. Previously, behavioral and molecular findings of lower propensity to despair behavior, anxiety and other signs of maladaptive stress response demonstrated in Tph2^+/− mice (Gorlova et al. 2020; Strekalova et al. 2021a, b; Svirin et al. 2022a) and in Tph2^−/− mice (Savelieva et al. 2008; Weidner et al. 2019) have been demonstrated. In this connection, it has been proposed that TPH2 deficiency might have not only negative consequences been associated with pathological aggression, impulsivity and other symptoms of deurodevelopmental pathologies, but also result to beneficial increase of stress resilience (Lesch et al., 2012).

Similarly to the present and previous findings, in a recent study of the PFC transcriptome in aggressive and non-aggressive cattle breeds, a number of upregulated transcripts were more than threefold higher in aggressive breeds (Kukekova et al. 2011; Eusebi et al. 2021). This may potentially mirror the fact that aggression, as a profoundly complex behavior, is accompanied by emotional and physical invigoration, thus requiring an involvement of enhanced neuronal activity (Kukekova et al. 2020). In light of this view, the downregulation of several Sox family members, such as Sox17 and Sox18, which represent transcription factors involved in diverse cellular events during development (She and Yang 2015), may suggest that these molecules mediate synaptic remodeling and neural network plasticity in aggressive animals. SOX transcription factors also impact tissue regeneration and act either as oncogenes or tumour suppressor genes, depending on the cellular environment. They are activated or inhibited through diverse genetic and epigenetic mechanisms, including DNA copy number variation, DNA methylation and abnormal microRNA expression (Grimm et al. 2020). Of note, altered expression of Sox17 in PFC was also reported in aggressive cattle breeds (Eusebi et al. 2021). Importantly, this transcription factor is thought to be implicated in many processes that go beyond the development stage. According to the study of Lange et al. (2014), SOX17 can regulate at least 267 genes. Other genes regulated by SOX17 may be more directly involved in controlling aggressive behavior, including Monoamine Oxidase A (MAOA) that regulates aggression across spices, since SOX17 binding sites were found in its promoter region (Chen et al. 2019). Thus, SOX17 is likely to be directly implicated in the mechanisms controlling aggressive behavior.

In line with the notion that SOX transcription factors regulate neurodevelopment, other transcriptional factors with similar functions were associated with excessive aggression in Tph2^+/−mutants. For instance, Ets1 is known to regulate a number of basic biological processes in normal cells and has been linked to the regulation of immune cell function playing roles in immunity and autoimmunity (Garrett-Sinha 2013; Dhara et al. 2022). Thus, besides potential functions in synapse formation and remodeling, this factor may be implicated in increased neuroinflammatory responses also found in models of stress-induced aggression (Svirin et al. 2022b, 2022). Similarly, Cdt1, encoding Cdc10-dependent transcript 1 protein, which was upregulated in the PFC of stressed Tph2^+/− mice, might be involved in neurogenesis and structural and functional circuit plasticity, as it is required for the first step in DNA replication, thus being critical for genome stability and to normal development (Pozo and Cook 2016). Likewise, Mybl2 (alias B-Myb) is a transcription factor of the MYB transcription factor family and a physiological regulator of cell cycle progression, cell differentiation and survival (Musa et al. 2017), is likely to be involved in neurogenesis and plasticity. The same holds true for Ruvbl2, involved in the remodeling of chromatin, DNA damage repair, and regulation of the cell cycle (Su et al. 2022), processes essential in neurogenesis and plasticity.

Remarkably, we found an upregulation of Zbtb20 in the stressed Tph2^+/− group. ZBTB20 was shown to be critical for the developing hippocampus. Increased methylation within the gene’s coding region was associated with major depressive disorder in genome-wide MeDIP-sequencing study on 50 monozygotic twin pairs that were discordant for depression (Davies et al. 2014). Given that depression and abnormal aggression are often comorbid, altered expression of Zbtb20 might underlie the behavioral abnormalities reported in stressed Tph2^+/− animals. Another noteworthy finding was the altered expression of the Hmga1 regulon in stressed mutant females, since a genetic lack of HMGA1 protein was reported to adversely affect insulin receptor (INSR) expression in cells and tissues in individuals with insulin resistance and type-2 diabetes. As a matter of fact, HMGA1 is considered as a downstream nuclear target of the insulin receptor signaling pathway (Foti et al. 2005; Chiefari et al. 2012). Restoration of HMGA1 protein expression enhanced INSR gene transcription and normalized cell-surface INSR levels and insulin-binding capacity (Foti et al. 2005). Our recent work evidenced abnormal INSR expression in stressed Tph2^+/− animals and aggressive mutants with genetic deficit of Gal5 gangliosides (Svirin et al., 2022b), that is keeping with the view of critical roles of INSR signaling in brain function (Pomytkin et al. 2018). Moreover, aberrant expression of Hmga1 was found to cause insulin resistance (Foti et al. 2005). While available literature is lacking any direct indications for a relationship between food deprivation, aggression and insulin receptor functions, a demonstration of compromised secretion of hypothalamic hormones in female mice altered insulin levels that are subjected to fasting (McCosh with et al. 2019) and other indirect findings (Pomytkin et al. 2018) additionally let to hypothesize the role of insulin receptor dysregulation via HMGA1-related mechanisms in pathological aggressive behavior. As such, changes in Hmga1 and thus INSR-mediated signaling might represent a mechanism underlying excessive aggression in stressed Tph2^+/− mice.

In addition, stressed Tph2^+/− mice displayed upregulated Nono regulon, whose protein product has been characterized as an transcriptional regulator implicated in intellectual disability in humans as well as cognitive and affective deficits in mice (Mircsof et al. 2015). The NONO protein was shown to moderate inhibitory synapse function, by regulating synaptic mRNA transcription and gephyrin scaffold structure, both representing critical mechanisms in the functional organization of GABAergic synapses and maintenance of excitatory/inhibitory balance. As GABAergic transmission is well documented to be involved in the regulation of aggression and GABA turnover was found to be altered in Tph2^+/− mice (Strekalova et al. 2021a, b), the changes in Nono are likely to be implicated in the behavioral phenotype of stressed Tph2^+/− animals. The link to the hormonal stress response between abnormal aggression and partial 5-HT deficiency in stressed female mutants may be mediated via upregulated Klf9, which is a key feed-forward regulator of the transcriptomic response to glucocorticoid receptor activity (Spoerl et al., 2012; Gans et al. 2020). Given the interrelationship between 5-HT system regulation and HPA axis, function changes in KLF9-regulated binding of glucocorticoid receptors may play a major role in the mechanisms of stress-induced aggression. Moreover, we detected downgulation of Atxn1 regulon, in which mutations cause spinocerebellar ataxia type 1, an inherited neurodegenerative disease characterized by a progressive loss of Purkinje neurons in the cerebellum. A link between altered Atxn1 expression and impulsivity as well as cognitive functioning was reported in ADHD patients (Rizzi et al. 2011) and associations between cognitive abilities and SNPs in ATXN1 showed replicated association, but only in patient cohorts ascertained for ADHD. It might be that upregulation of Atxn1 in the present work supports its role in impulsive behavior that overlaps with aggression. Finally, genome-wide association studies revealed a link between expression of Notch1, that was downregulated in stressed Tph2^+/− mutants, and the incidence of epilepsy (Huan et al. 2017) and ADHD (Alemany et al. 2016) that again might be interpreted as a manifestation of dysregulated brain inhibitory processes, a mechanism closely related to excessive aggression. Downregulation of Rbpj, acting as a transcriptional activator of Notch target genes further supports the involvement of Notch1-related mechanisms in abnormal aggression.

Together, within the upregulated enriched regulons in the aggressive mouse cohort, we found biological functions associated with processes, such as cellular functions, brain development and plasticity, immune responses, insulin receptor-mediated functions, GABA-related mechanisms of impulse control as well as links to neurological, neurodevelopmental and psychiatric disorders. In translational context, our findings are in line with clinical evidence that link the Tph2 gene polymorphisms, e.g., G703T to anger-related traits and the expression of anger in women (Yang et al. 2010). Other variants of the Tph2 gene were also associated with a higher incidence of anxiety disorder in women and with peripartum major depression (Lin et al. 2009; Fasching et al. 2012).

Taken together, our findings reveal new molecular signatures underlying the pathological aggression of female Tph2^+/− mice exposed to stress and food deprivation. Further studies are required to shed light on the detailed mechanisms of the relationships between 5-HT deficiency and predation stress and female aggression in the context of gene × environment interaction. Our study supports the use of this mouse model of female aggression to investigate gene x environment interaction in the regulation of emotional behavior in the context of neurodevelopmental and psychiatric disorders.