Abstract
The cumulative load of genetic predisposition, early life adversity (ELA) and lifestyle shapes the prevalence of psychiatric disorders. Single nucleotide polymorphisms (SNPs) in the human FKBP5 gene were shown to modulate disease risk. To enable investigation of disease-related SNPs in behaviorally relevant context, we generated humanized mouse lines carrying either the risk (AT) or the resiliency (CG) allele of the rs1360780 locus and exposed litters of these mice to maternal separation. Behavioral and physiological aspects of their adult stress responsiveness displayed interactions of genotype, early life condition and sex. In humanized females carrying the CG-but not the AT-allele, ELA led to altered HPA-axis functioning, exploratory behavior and sociability. These changes correlated with differential expression of genes in the hypothalamus, where synaptic transmission, metabolism, and circadian entrainment pathways were deregulated. Our data suggest an integrative role of FKBP5 in shaping the sex-specific outcome of ELA in adulthood.
1 Introduction
Stress responses are essential to adjust physiology and behaviour to recurrently changing environmental demands [1], but corrupted stress responses are a hallmark feature of psychiatric conditions [2]. The susceptibility or resilience to develop psychiatric disorders can be attributed to interactions of genetic predispositions and environmental factors [3]. Among environmental factors, early life adversity (ELA) is found to be especially detrimental given that aberrations during development will influence the affected individuals throughout life [4]. Childhood maltreatment is common in the history of many psychiatric patients and comprises experiences of physical, sexual and emotional abuse, as well as physical and emotional neglect [5]. Such experiences during development shape disease prevalence in later life through alterations in HPA-axis programming, stress coping strategies and brain connectivity [6].
With respect to genetic predispositions, the regulation of glucocorticoid signaling is a prominent research target since glucocorticoids are a key messenger for the spread and initiation of stress responsive signaling. This regulation is fine-tuned in a timing- and dose-dependent manner and depends on the individual cellular set-up such as the relative expression of glucocorticoid receptors and its regulators [7]. Expression levels of FKBP5, a potent negative regulator of glucocorticoid signaling, is part of this cellular identity and is itself a target of glucocorticoid-mediated gene transcription [8]. Single nucleotide polymorphisms (SNPs) inside the human FKBP5 gene are associated with differential induction of the FKBP51 protein upon glucocorticoid stimulation [9] and add to the variability of stress perception and response in the population [10]. Carriers of the high induction allele rs1360780-A/T of FKBP5 who suffered from ELA are more prone to develop psychiatric symptoms in later life than individuals without such preconditioning [11]. Importantly, sex-dependent differences in the interaction of FKBP5 and life adversities have been associated to a higher prevalence of depression in females [12]. Despite the strong negative impact of psychiatric disorders on quality of life and productivity, the underlying processes linking FKBP5 genotypes, stress regulation and pathological transitions are not fully understood. Animal models offer a possibility to investigate gene × environment interactions in a timely resolved manner. In depth analyses of laboratory mouse sequences in-house indicated numerous Fkbp5 SNPs that vary by strain. However, no SNPs at the same location or with the same functional impact as found in humans occur naturally in rodents.
This lack of an animal model suited to exploring human FKBP5 SNPs hinders elucidation of causal relationships and mechanisms underlying disease development and progression. Therefore, we previously generated Fkbp5-humanized mice carrying either the risk-associated high induction AT-allele of rs1360780 or the resiliency-associated CG-allele. Initial characterization of primary CNS-cell types derived from these mice revealed that the presence of the AT-allele results in the increased expression of Fkbp5 upon stimulation of the glucocorticoid receptor compared to the CG-allele [7]. This initial characterization prompted us to exploit this new model to examine the Fkbp5 × ELA interactions on the stress response system in adulthood. We exposed AT and CG-allele carrying mice to prolonged maternal separation stress, since this paradigm is broadly used to mimic ELA in rodents [13]. When mice reached adulthood, the performance of the HPA-axis and behavioral response of Fkbp5-humanized mice to mild stressors were measured. Furthermore, we investigated the transcriptomic profiles in several brain regions engaged in stress processing. Lastly, astrocytes and neurons derived from human induced pluripotent stem cells (hiPSCs) were analyzed for SNP-based differens in their expression profiles.
The goals of the study were to validate the Fkbp5 × ELA model by 1) determining wether ELA would cause alterations in the offspring’s adult behaviour and physiology compared to controls, 2) determining whether risk AT-allele carriers would respond differently to ELA than CG-allele carriers, 3) assessing which pathways are involved in the adaptation to ELA in context of risk and resilience associated SNPs. A more far-reaching aim was to demonstrate that the humanized Fkpb5 × ELA mouse model can be used to further investigate the influence of the human FKBP5 gene variants on the risk and resilience to stress and to further elucidate their contribution to psychiatric disorders.
2 Results
Prolonged separation from mothers and peers was performed for the first three weeks of postnatal life to model ELA. In parallel, control mice were housed with littermates and received undisturbed maternal care until weaning. An overview of the group sizes of the cohort is provided in Tab. 1. On postnatal day 21, pups were weaned and grown to adulthood with physiological and behavioral examination starting at 10 weeks of age (Fig. 1). Exploration of novel environments offers an easily accessible measure of mild stress in rodents [14]. Therefore, we challenged control and ELA-exposed mice with novel situations to probe for their stress coping strategies. The same procedures were simultaneously carried out in wild type mice of both sexes. The data on their HPA-axis functioning (Fig. S1) and behavior (Fig. S2) are visualized in the supplements for reference. Statistical analyses were performed jointly for males and females to address differences between sex, ELA exposure, Fkbp5-genotypes and the interaction thereof. Details of the descriptive analyses, model summaries and analysis of variance (ANOVA) results are provided in the supplements 7. A significant effect of sex × genotype × treatment interaction and significant two-way interactions in the vast majority of measured parameters were detected and are detailed in the following paragraphs.
2.1 Early Life Adversity and Fkbp5-Variants Shift and Attenuate Diurnal HPA-Axis Rhythmicity
To measure the impact of Fkbp5 SNPs in combination with ELA on the diurnal performance of the HPA-axis, the plasma corticosterone concentration of samples collected at three time points was assessed. As confirmed in the wild type mice (Fig. S1), these timepoints were reflecting the diurnal nadir (morning), peak (evening) and one intermediate state (noon). In control females carrying the CG-allele, the expected increase of plasma corticosterone over the course of the day was observed, with a clear peak towards the evening (Fig. 2a, Tab. S1, Tab. S2, Tab. S3). Following ELA exposure, the highest concentration was instead measured at noon. The increase of plasma corticosterone levels in AT-allele carrying control females was not statistically significant, regardless of ELA exposure.
In line with these findings, the adrenal weight was increased by ELA in CG-allele carrying females, while in the AT-allele carrying females, the adrenal weight tended to be already increased in control mice compared to CG controls without further increase upon ELA (Fig. 2b, Tab. S4, Tab. S5, Tab. S6).
In Fkbp5-humanized males, the diurnal plasma corticosterone concentration peaked towards noon with CG-vs-AT-allele carriers showing a decrease towards the evening, regardless of ELA exposure (Fig. 2c). The detected diurnal amplitude of corticosterone was smaller in males than females. No significant differences in the adrenal weights were observed among males (Fig. 2d), but male vs. female adrenal weights were significantly lower.
Taken together, female AT-vs. CG-allele carriers are genetically predisposed to less pronounced diurnal HPA-axis rhythmicity resulting in elevated corticosterone levels at time points were mice usually would rest. Lower diurnal corticosterone amplitudes and adrenal weights in males vs. females suggest a different corticosterone secretion capacity between sexes.
2.2 Early Life Adversity Increases Responsiveness to Novel Environments Dependent on Fkbp5-Genotype and Sex
Exposure to novel environments as mild stress was applied to determine natural behavior and coping strategies. First, behavior in open field test arenas was assessed to obtain a measure of locomotor activity at the beginning of and throughout the murine active phase (18:30 – 05:30). Overall activity within the first 15 minutes, including running and rearing, was assessed by measuring the frequency of crossing light beams (Fig. 3a, Tab. S7, Tab. S8, Tab. S9). During this period, the activity decreased over time with early life condition and sex showing an interaction with time. As in wild type females (Fig. S2a), the group of CG control females displayed habituation in the shape of a strong decrease in activity, while the exposure to ELA led to flattening of the 15 minutes activity profile and thus slower habituation (Fig. 3a). Habituation in AT-allele carrying females tended to be slower than in CG-controls, regardless of early life condition.
Analyses of the total nocturnal distance revealed main effects of sex and early life condition, as well as an interaction effect of ELA × genotype (Fig. 3b, Tab. S10, Tab. S11, Tab. S12). While ELA-exposed female CG-allele carriers were more active than controls, AT-allele carrying females with ELA experience were indistinguishable from controls.
In the spontaneous alternations T-maze, ELA did not affect the fraction of alternations between left or right side of the maze, irrespective of genotype or sex (Fig. S3a, Tab. S14, Tab.S13, Tab. S15), suggesting no impact on working memory performance. However, ELA-exposed mice performed the task significantly faster than the respective control group and females were quicker than males (Fig. S3b, Tab.S16, Tab. S17, Tab. S18).
In the dark-light test, ELA decreased the mean time spent in the dark compartment. Moreover, a trend for Fkbp5-genotype related effects was seen, with CG-allele carrying females compared to controls showing the ELA effect, while in the AT-allele carriers the control group was indistinguishable from ELA-exposed females (Fig. 3c, Tab. S19, Tab. S20, Tab. S21).
Finally, we measured social preference in the 3-social-chamber test. Pairwise comparisons of compartment effects separated by early life conditions, genotype and sex revealed significant differences: CG-allele control females showed social preference, measured by the time the mouse spent in the nearest vicinity of the cylinder with the social stimulus (Fig. 3d, Tab. S22, Tab. S23, Tab. S24). The exposure to ELA led to decrease of this parameter, while simultaneously we observed a significant increase in the time spent in the chamber, but in 5 cm distance from the occupied cylinder (Fig. 3e, Tab. S25, Tab. S26, Tab. S27). In contrast, AT-allele carrying control females spent less time interacting with the unfamiliar mouse, as compared to CG-allele carrying controls. ELA did not further change this parameter, and the time of direct interaction vs. time in ‘social distance’ was similar in the AT-allele carrying controls and ELA-exposed females.
In males, the activity measured in the open field arena (Fig. 3f and g) and working memory assessed as spontaneous alternations in the T-maze (Fig. S3a) were similar among groups. Like in females, male CG-allele carriers with ELA tended to spend less time in the dark compartment (Fig. 3h), and to complete the T-maze test faster (Fig. S3b) than controls. In contrast to females, the social preference was not affected by ELA in male CG-allele carriers, but decreased in ELA-exposed AT-allele carrying males (Fig. 3i, Fig. 3j).
Overall, the data on behavioral responses to mild stress elicited by novel environments suggest that the effects of ELA on these read outs depend on the genetic variants of Fkbp5 × sex.
2.3 HPA-Axis Responses are Stronger in Females than Males
To probe the HPA-axis reactivity to acute induction and negative feedback, we measured plasma corticosterone after 5 minutes of restraint stress and 6 h after a single intraperitoneal injection of the synthetic glucocorticoid dexamethasone. In all mice, corticosterone increased in response to restraint stress without a differential effect of genotype. However, the slope was steeper in females (Fig. 4a) compared to males (Fig. 4b) among ELA-exposed mice. Similarly, all mice responded to dexamethasone with reduced corticosterone levels, suggesting a suppression of the endogenous corticosterone secretion. Post hoc analyses revealed that the slope of decrease was overall steeper in females exposed to ELA relative to controls (Fig. 4c). In males, no effect of ELA or Fkbp5-genotype on the HPA-axis responsiveness to negative feedback was statistically significant (Fig. 4d).
In summary, the responsiveness of the HPA-axis is preserved in Fkbp5-humanized mice.
2.4 Transcription in Stress-Responsive Brain Regions is Affected by Fkbp5 × ELA
To identify transcriptional differences that could be related to the observed differences in behavior and HPA-axis physiology of Fkbp5-humanized mice × ELA, mRNA sequencing and analyses of differential gene expression were carried out. Given the sexual dimorphism in the in vivo experiments, analyses were limited to females and focused on hypothalamus, ventral and dorsal hippocampus as brain regions engaged in stress regulation [15].
In the SNP-comparison among controls, more differentially expressed genes (DEGs) were found in the hypothalamus (579), followed by ventral (41) and dorsal (2) hippocampus (Tab. 2). Among ELA-exposed individuals, substantially more DEGs between the SNP variants were detected than in controls, underscoring the interaction of ELA × Fkbp5-genotype. Looking at the effect of ELA, fewer differences were detected in AT-allele carriers (114) than in CG-allele (903) carriers. This matches to the behavior and HPA-axis data, where the AT-allele alone was found to predispose to a ‘stress-like’ phenotype, with few additional impact of ELA.
Adopting knowledge from the SNP effects in humans, the overlap and uniqueness of the DEGs from the two comparisons and subgroups were analyzed for nomination of potential resiliency- or vulnerability-related genes. Genes linked to CNS-development such as Mab21l2, Gart and Lipt2 were spotted as potentially vulnerability-related and were changed in opposite directions, with AT- vs. CG-allele carriers displaying a lower expression.
A second analysis focussing on gene clusters related to (developmental) neurological disorders using a two-step core and comparison analysis of the commercial software Ingenuity (Qiagen) confirmed that the ELA-responsive DEGs in both mouse lines have an impact on neurological and psychiatric symptoms (Fig. S4). In eight of the shown 30 deregulated clusters e.g. comprising ‘congenital neurological disorder’ or ‘learning’, the effects were opposite between AT- vs. CG-allele carriers.
In sum, the counts of DEGs and their accordant vs. discordant overlap between the analyzed subgroups suggest that the Fkbp5 × ELA interaction on gene expression may have relevance for neurologic and psychiatric symptomatology.
2.5 The AT-Allele and ELA Reduce CNS Communication but Increase Metabolism
To identify how the DEGs might be linked to disorders via their role in cellular pathways, their over-representation in metabolism and signaling-related pathways listed in the Kyoto Encyclopedia of Genes and Genomes (KEGG) was assessed. The analyses revealed significantly altered pathways in the hypothalamus and ventral hippocampus (Tab. 3). The direction of change between Fkbp5-genotypes differed dependent on function, with pathways related to neuronal communication rather showing a downregulation, and pathways related to metabolism rather showing an upregulation in AT- vs. CG-allele carriers.
In the hypothalamus, the most significantly downregulated pathways included circadian entrainment, regulation of synaptic plasticity via long-term potentiation and depression as well as activity of dopaminergic and cholinergic synapses together with changes in calcium, cAMP and oxytocin signaling. In the ventral hippocampus, reduced expression of synaptic communication in AT- vs. CG-allele carriers was repeated. Especially in the ELA-subgroup, lower expression of genes related to cAMP signaling and dopaminergic synapses were found in AT-allele carriers compared to CG-allele carriers. Independent of strain, ELA was linked to lower expression of transcripts related to endocannabinoid and circadian entrainment relative to controls. For genes in pathways related to metabolism, such as protein absorption and digestion in the hypothalamus or ribosome activity and oxidative phosphorylation in the ventral hippocampus of controls, higher expression in AT-allele carriers relative to CG-allele carriers was observed.
The mRNA of neurons and astrocytes derived from human induced pluripotent stem cells (hiPSCs) of rs1360780 SNP carriers was sequenced and used to qualitatively validate the SNP-dependence of the observed differences in an independent expression system. In both cell types, comparable SNP-based expression differences, which might indicate less synaptic communication in AT- vs. CG-allele carriers, were seen. However, the distribution within the pathways differed between hiPSC- and mouse derived samples. For example, more differential expression in the upstream vs. downstream members of the circadian entrainment pathway was seen in the Fkbp5-humanized mice, while in the hiPSCs rather the expression of downstream targets was changed (Fig. S5). Moreover, the expression patterns in astrocytes vs. neurons were more similar to the patterns seen in mice.
The KEGG pathway analyses imply that ELA and the AT-allele both lead to less entrainment of diurnal HPA-axis rhythmicity. This lower entrainment of sleep-wake states may interact with the decreased ability of AT- vs. CG-allele carriers to process incoming inputs via synaptic communication.
2.6 Lower Glucocorticoid Sensitivity of the Hippocampus is Modulated by Fkbp5 Genotype
To estimate how much impact the potentially altered glucocorticoid exposure due to differences in circadian entrainment and synaptic signaling might exert on the hypothalamus, ventral and dorsal hippocampus, the expression levels of genes related to glucocorticoid signaling were compared (Fig. 5). This analysis provides insights in the likelihood of the brain regions to respond to glucocorticoid stimulation. While expression levels of the glucocorticoid receptor (Nr3c1) and heat shock protein 90 (Hsp90ab1) were comparable between all three brain regions, the mineralocorticoid receptor (Nr3c2) was less expressed in the hypothalamus than in the hippocampus, with the ventral hippocampus displaying the highest expression. Moreover, Fkbp5 was less expressed in the hypothalamus than hippocampus and the AT- vs. GC-allele was associated with a lower Fkbp5 expression in dorsal and ventral hippocampus. Considering the gene functions, the hypothalamus appears to be more sensitive to glucocorticoid receptor mediated signaling than the hippocampus, with CG- vs. AT-allele hippocampi being more protected. The decreased cerebral expression of genes related to synaptic communication in AT- vs. CG-allele carriers might be a compensatory mechanism to prevent excessive excitation. To test whether the expression levels of the identified DEGs and Fkbp5 could be linked to the observed behavioral and physiological differences, tissue-wise correlation analyses were carried out. For each brain region, the top 10 correlations are provided in Tab. S34 (the full list of correlations will be provided upon request). In all three brain regions, the majority of DEGs correlated with Fkbp5. In the hypothalamus, gap junction protein β 1 (Gjb1) showed a correlation with the time spent in the dark compartment of the test arena, while the membrane-associated, tyrosine-specific kinase 1 (Pkmyt1) and the nicotinic acetylcholine receptor subunit 7 (Chrna7, regression shown in S6) were linked to morning corticosterone levels. This could indicate an association between some hypothalamic DEGs and differences in HPA-axis functioning and behavior. The correlation analyses suggest a linkage between expression levels of Fkbp5 and DEGs in brain regions relevant for stress processing.
3 Discussion
The present study has demonstrated a gene × environment interaction in novel Fkbp5-humanized mice indicating that the model is suited to investigate the effects of ELA in the context of risk- and resiliency-related SNPs. Early life adversity elicited by maternal separation has differential impact on adult physiology and behavior based on genetic predisposition imparted by Fkbp5 alleles. This is demonstrated by changes in locomotor, social, and anxious behavior. Additionally, diurnal corticosterone rhymicity is moderately altered as seen at a functional level via HPA-profiling and on molecular levels through altered gene expression in the circadian entrainment pathway. Differential gene expression in brain regions relevant to stress regulation shows an enrichment for pathways linked to neural communication and brain disorders. Many of the differentially expressed genes are correlated with Fkbp5 levels. In the tests utilized here, the impact of Fkbp5 SNPs and ELA was greater in females than males.
These stronger effects of Fkbp5 × ELA in female compared to male mice match previously reported sexual dimorphism in responsiveness to ELA in animals [16] and was discussed in humans [17]. Notably, ELA and sex hormones both influence maturation kinetics and thus the development of cerebral regions implicated in glucocorticoid regulation [18]. The interactions of the SNP rs1360780, sex and ELA observed in the mice presented here and in humans [11, 12] could thus be explained by the regulatory capacity of Fkbp5 on glucocorticoid signaling. Based on sex-dependent correlations between FKBP5 levels and depression and anxiety scores as well as with nadir cortisol levels, FKBP5 was suggested as a female-specific biomarker for prolongued cortisol load and the associated risk of psychiatric disorders [19]. In line with this correlation, we observed associations between genotype and nadir corticosterone levels in Fkbp5-humanized mice, with AT-allele carrying females displaying higher morning glucocorticoid levels than CG-allele carrying females. The sexual dimorphism in the effect of ELA indicate that the novel Fkbp5-humanized mouse model offers the possibility to further investigate the networking of ELA, sex and disease-related SNPs.
In addition, the data provide mechanistic insights into how Fkbp5 SNPs may contribute to the shaping of overall physiology and the stress response system. As negative modulator of glucocorticoid receptor maturation, Fkbp5 holds the potential to inhibit glucocortocid signaling. At the same time, its expression depends on recent glucocorticoid exposure since Fkbp5 itself harbors glucocorticoid response elements [8]. The higher induction of the AT-allele in CNS cell types of Fkbp5-humanized mice upon glucocorticoid stimulation could thus be expected to result in stronger or longer inhibition of subsequent glucocorticoid signaling [7]. In vivo, this stronger induction of inhibitory potential via Fkbp5 in AT-allele carriers could lead to dampened negative feedback to the HPA-axis and a prolonged interval of elevated glucocorticoid levels. The negative feedback loop is furthermore critical for the maintenance of oscillation patterns and function [20]. The reduction in the complexity of ultradian fluctuation and the resulting decreased variability of HPA-axis reactivity in AT-allele carriers could decrease their flexibility to respond to novel environments. Behavioral evidence of this differential responsiveness could include the alterations in light-dark box testing, locomotor habituation, and abnormal social behavior as seen in this study. In humans, differences in HPA-axis responsivess to environmental stimuli, e.g. in the Trier Social Stress Test, between human AT- and CG-allele carriers has been demonstrated [10]. The findings imply that Fkbp5 genotype dependent regulation of ultradian HPA-axis activity might be a core molecular mechanism that contributes to the variability seen in human stress responsiveness, which ultimately plays a role in distinction between healthy adaptation or pathological alteration in the aftermath of stress [21].
Another environmental stimulus that can affect glucocorticoid rhythms is the lightdark cycle [22]. One commonly investigated manifestation of this circadian rhythmicity is the pronounced increase of glucocorticoids prior to awakening [23]. Mechanistically, the ability to detect light in the retinal ganglia and to signal this via the suprachiasmatic nucleus to the periphery is a crucial trigger for the awakening response [24]. In AT- vs. CG-allele carriers, flatter diurnal glucocorticoid profiles were paralelled by lower expression of circadian entrainment related genes even though histological analyses of the eyes (data not shown) indicated no differences in the ability to detect light. This underscores the relevance of self-maintaining feedforward and feedback loops in regulating overall physiology throughout the day. While external light signals can synchronize individuals to a 24 h cycle [25], the internal gene expression driven clock seems to define the shape of the circadian glucocorticoid profile and thus when and how strong individuals are likely to respond to challenges. In humans, modulation of the cortisol awakening response was reported to influence their performance during the upcoming day and was dependent on the anticipation of challenges [26]. The awakening response is used clinically to identify individuals with certain personality traits that are vulnerable to develop psychiatric disorders [27], and for the diagnosis of depression [28].
Besides impaired awakening responses, differences in kinetic and responsiveness of the HPA-axis, e.g. to acute stress or dexamethasone exposure, between psychiatric patients and healthy controls have been demonstrated [29]. In the present study, no dysfunction of HPA-axis responsiveness was observed, which indicates that the combination of ELA and genetic predisposition via the AT-allele of Fkbp5 alone might not be sufficient to cause full pathology. This is in agreement with the Research Domain Criteria framework proposing a continuum between ‘normal’ and ‘pathological’ which needs to be better understood in order to alleviate symptoms. Accordingly, the transition to pathology occurs over a life time and is a multidimensional process shaped by numerous genetic and environmental factors that introduce subtle changes which jointly alter networking of physiological systems [30]. As in humans, the Fkbp5-humanized mouse model demonstrates changes in basal HPA-axis activity dependent on genotype and early life experience, with more prominent effects in females than males. These alterations in non-stimulated HPA-axis functioning were suggested to have an impact on sleep-wake states, responsiveness to environmental stimuli and vice versa [31]. In the long run, insufficient adaptation could contribute to allostatic load and finally development of disorders [21]. However, the cumulative stress load in this study was low since the animals were not exposed to any severe or chronic stressors during later life.
Nevertheless, the Fkbp5 × ELA model shows indications of changes in the psycho-immune-neuro-endocrine system that are commonly seen in response to chronic stress. Reduced expression of immediate early genes as markers of plasticity in the prefrontal cortex and hippocampus as well as elevated mitochondrial respiration in response to repeated mild stress during adulthood was previously reported [32]. In the present study, the increased expression of genes related to oxidative phosphorylation in the hippocampus of AT- vs. CG-allele carriers is an interesting parallel, as is the reduction of genes related to synaptic communication. Reduced neural communication and plasticity might become maladaptive since dendritic retraction has been described to render the hippocampus more vulnerable to neurotoxic or metabolic challenges [33, 34]. The longer the time window of decreased plasticity and increased vulnerability exists, the higher is the likelihood of a co-incidential high metabolic demand. Stressful situations only transiently elevate energetic demands while simultaneously decreasing the neuronal supply with glucose [35]. Unique stress events may thus not cause irreversible harm to the hippocampus, and AT-allele carriers might even benefit from their inherent higher expression of mitochondrial genes. Under prolonged exposure to glucocorticoids, increased oxidative phosphorylation in AT-allele carriers might produce excessive amounts of neurotoxic reactive oxygen species which may damage the hippocampus. Findings of this study imply more glucocorticoid signaling in the hippocampus of AT-relative to CG-allele carriers since the glucocorticoid signaling inhibitor Fkbp5 had a lower expression level while nadir corticosterone levels were increased in female AT- vs. CG-allele carriers. Cumulatively, this mechanism could contribute to the loss of hippocampal volume in stress-related disorders such as depression and would explain why AT-allele carriers are more prone to develop disorders than CG-allele carriers [36]. The proposed sequence of alterations on cellular and circuitry level from healthy to allostatic load and allostatic overload conditions is outlined in Fig. 6. Assessment of behavior and physiologic read outs in Fkbp5-humanized mice that experienced both, ELA and more severe or chronic stress paradigms, would resolve these questions.
Moreover, the combination of Fkbp5-SNPs and ELA with simultaneous or sequential stress hits could enable prediction of and intervention at critical transition points during the development and progression of psychiatric symptoms.
4 Conclusion
The cumulative load of genetic predisposition, unfavorable environmental influences during development and repeated exposure to stressful events increases the prevalence of psychiatric disorders in affected individuals. The glucocorticoid-induced expression of Fkbp5 is a hub for integrating lifetime and recent stressful experiences. Simultaneously, Fkbp5 modulates responsiveness to acute stressors as negative modulator of glucocorticoid signaling. The naturally occurring Fkbp5-SNPs in laboratory rodents do not feature comparable functional effects as rs1360780 in humans, where the AT- vs. CG-allele is more strongly induced by glucocorticoids and linked to the etiology of psychiatric disorders. To enable studying in more detail the mechanistic impact of the human SNP on stress physiology and the etiology of psychiatric disorders, Fkbp5-humanized mouse lines carrying either the AT- or CG-allele of this SNP were generated. Characterization of the Fkbp5 × ELA mouse model showed mechanistic and face validity with aspects of psychiatric disorders. Female AT- vs. CG-allele carriers after ELA showed attenuated diurnal rhythmicity of glucocorticoids, lower activity, and less responsiveness to novel environments. On a molecular level, reduced expression of genes related to circadian entrainment and synaptic communication as well as increased expression of genes related to mitochondrial respiration between AT- vs. CG-allele carriers imply a genetic predisposition of their psycho-immune-neuro-endocrine system to allostatic changes reported in a mild chronic stress settings. Since ELA lead to decreased circadian entrainment in the hippocampus, which in turn influences the circadian entrainment in the hypothalamus, the combination of ELA and Fkbp5 SNPs could synergistically modify the HPA-axis to respond less to stimuli. Given that dynamic variability in glucocorticoid levels and plasticity are required for adaptation to challenges, this predisposition increases the risk of an unsuccessful resolution of allostatic loads and thus elevates the risk of developing stress-related disorders. In combination with severe or chronic stress exposure, the observed Fkbp5 × ELA interactions likely contribute to the etiology of stress-related pathology. Taken together, we are confident that this novel animal model will contribute to more comprehensive analyses of FKBP5-induced alterations in the stress response network that causally lead to the development of pathology.
5 Methods
5.1 Generation of Transgenic Mice
Two novel transgenic mouse models were created by and are publicly available at Taconic Biosciences, carrying either the cytosine (C)/guanine (G) variant at position 3622 in the human FKBP5 gene (C57BL/6NTac-Fkbp54571 (FKBP5) Tac) or the high-induction adenine (A)/thymine (T) version of rs1360780 (C57BL/6NTac-Fkbp5tm4570 (FKBP5) Tac). In short, the murine Fkbp5 was exchanged with the human FKBP5 coding region, keeping the 3’ and 5’ UTR of the mouse. Full details of the method are available at [7]. Homozygote mice were bred in-house to be used in the experiments.
5.2 Animal Husbandry
Standard laboratory conditions were adhered to (20-24°C, 45-55% humidity, 12-hour light/dark cycles (sunrise 6:00, sunset 18:00), ad libitum access for standard laboratory chow and water, enrichment (wooden block, red plastic shelter house and tube and paper stripes) provided). Mice had not been used in any other study prior to sacrifice. All animal experiments were performed under allowance of the regional council for animal welfare (Regierungspräsidium Tübingen, Baden-Württemberg, Germany, licence VVH 17-009) and in compliance with directive 2020/63/EU and ARRIVE guidelines. An overview of the study cohort is provided in Tab. 1. Sample size estimation was based on empirical knowledge regarding measurement accuracy or reproduction errors of in vivo and ex vivo methods (noise) and the expected effect sizes. Since we are the first to characterize these novel transgenic mice, no data was available for exact effect size estimations. Instead, published and in-house observed effect sizes of stress manipulations that were deemed to be biologically relevant were used as reference. As result of these considerations and based on our previous experience with animal experiments, n = 8 was agreed upon to be a suitable trade off between power for effect detection and the amount of animals required. Scheduled mating was used for breeding of the animals. On the day of birth, the litters were assigned to control or maternal separation in a way that group sizes were balanced as far as possible. Sex balance and equal litter sizes were not enforced since culling of littermates / offspring would introduce counfounding stressors and unnecessary suffering. As a result and since litter sizes and sex ratio within litters are non deterministic, final group sizes varied. After testing for equivalence, the WTs of the AT- and CG-allele strains were pooled which resulted in double the number of the transgenic groups.
5.3 Maternal Separation
Separation from mothers and litter mates at different times of the day for three hours starting from post-natal day two until 21 was carried out. During separation, heating pads were placed below the cages to keep the pups warm despite the lack of nesting material and low amount of saw dust. Mothers were kept at the other side of the room in their home cages. Maternal behaviour during separation and after reunion was observed to decrease after the first few separation sessions. Blinding of the experimenter was not possible given these obvious manipulations. After weaning on post-natal day 21, adolescent mice were group-housed until an age of 6 weeks. Afterwards, if not required earlier due to aggressive behavior, animals were single housing.
5.4 Behavioral Test Battery
Locomotion, dark-light preference, sociability and spatial working memory was assessed after mice were grown up to young adults. An overview of the timelines of the behavioral test battery is provided in Fig. 1. Within measurement sessions, mouse strains were mixed but controls and ELA-exposed mice as well as males and females were separated to avoid olfactory or auditory cues being transferred between groups, introducing distress or confounding effects. The experiments were first performed in males (controls, then ELA-exposed) and subsequently in females. Assignment to arenas or order of measurements within the day was randomized and for the social chamber and T-maze test experimenters were blinded regarding test groups. Arenas were extensively cleaned between each measurement and between groups. One day pause was kept to eliminate potentially distracting olfactory cues from the room.
5.4.1 Open Field Test
A maximum of 30 arenas of 45 by 45 cm size were evenly illuminated with 267 lx, water gel and food pellet were placed around the borders of the center zone and one handful of saw dust from each mouse’ home cage were distributed inside the arena. Mice were brought to the measurement room at least two hours before the session for recovery from transport and habituation to the room. Shortly before 17:00, the actimot (TSE Systems, Bad Homburg, Germany), which detects movement of animals via breaking of light beams in x, y, z direction, was switched on, mice were placed in the middle of the measurement chambers and their locomotion was recorded over night until 07:00 to obtain measures of their activity in the light and dark phase (lights off 18:00, lights on 06:00).
5.4.2 Dark Light Test
Measurement of preference for the dark or light compartment were also performed with the actimot system, while a 1/3 of the arena was kept dark (2.2 lx) using a black plastic house with a circular door to allow mice to freely travel between both compartments. At beginning of the measurement, mice were placed in the dark compartment facing the corner away from the door. Experiments were performed between 08:00 to 11:00 in the morning and lasted for 30 minutes.
5.4.3 Three Social Chamber Test
Sociability was tested in an arena divided into three compartments of equal size (60 × 40 × 22 cm, Ugo Basile) with sliding doors between the compartments. The left and right section contained a mesh cylinder (7 cm in diameter, 15 cm height). Mice were habituated to the measurement room 1 hour prior testing and experiments were performed in the morning from 7:00-12:00. The arena was evenly illuminated with 23 lx. After recording the 5 minutes habituation phase of the tested mouse to the arena, an unfamiliar stimulus mouse was placed in one of the cylinders. Choice of side was evenly distributed across groups. Stimulus mice were juvenile, of the same sex as the tested mouse, habituated to the cylinder and used twice per day with 1-hour break between measurements. Behavior of the tested mouse was videotaped for 10 minutes and analyzed using an automated tracking software (TopScan CleverSys Inc., USA). Main readouts were the time spent, the amount of entries, the activity during the visit measured by distance and speed, the latency to first enter and the latency to end the first visit. This was assessed for the chamber as a whole and a zone surrounding the cylinders. As secondary readouts, locomotion and immobility were measured.
5.4.4 Alternations T-Maze
Mice were moved to the testing room the day before their performance in the T-maze was assessed. At beginning of the measurement, the mouse was placed in the starting box for 5 seconds before the door to the arena was opened. In the first trial, the animal was either forced to enter the right or the left arm by closing the door to the respective other arm. Starting sides were evenly distributed across sexes, early life conditions and mouse lines. Every mouse was tested once and had to complete 14 trials consisting of entering one arm, closing the door to the non-chosen arm, returning of the animal to the starting zone and opening all passages to enable free choice of side for the next trial. The maximal allowed duration was set to 14 minutes and if an animal completed less than 7 trials it was excluded from further analysis. This was the case for 1 CG-allele carrying male control. Dimensions of the arena were 20 cm height, 8.5 cm corridor width, 30 cm lengths of each arm, 54 cm length of the starting zone. The test room was illuminated with 230 lx while above the T-maze light intensity was set to 50 lx. To enable spatial discrimination, navigation objects with differing shape and color were placed outside of the left and right arm of the arena.
5.5 HPA-Axis Performance
Blood was sampled from the vena saphena by immobilizing the mouse (Broome Rodent Restrainers, Harvard Apparatus, Cat.No.52-0460, MA 01746, USA), shaving and anointing the left leg and stinging into the vein with a lancet (Solofix, B. Braun, Cat.No. 6182003, Melsungen, Germany). Blood droplets were collected in K2-EDTA-containing capillaries (Microvette, Sarstedt, Cat.No.16.444, Nümbrecht, Germany) and stored on ice prior to centrifugation at 20000 g for 20 min at 4°C. The whole process from cage opening to collection of the last drop was carried out within less than 1.5 minutes to avoid a procedure-associated rise of corticosterone [37]. Plasma aliquots were frozen immediately.
For tracking of basal diurnal rhythmicity, blood was collected in the morning (06:45 – 07:15), afternoon (12:45 – 13:15) and evening (18:45 – 19:15). To obtain a measure of plasma corticosterone levels after stress, on another day at about the same time when the morning blood sample was drawn, mice were kept inside the restrainer for 5 minutes before puncturing the vein. Negative feedback to the HPA-axis was investigated by comparing rise or fall of plasma corticosterone levels between morning and 6 hours after injection of saline (NaCl 0.9%, B. Braun, Cat.No. FREU950) or with 0.001 mg/kg dexamethasone (DexaHexal 4 mg/ml diluted in saline, Hexal, Holzkirchen, Germany) two days after saline injection. An overview of blood sampling time points is provided in Fig. 1. Concentrations of corticosterone were quantified using an enzyme-linked immunosorbent assay (DetectX Corticosterone Enzyme Immunoassay Kit, Cat.No. CEA540Ge, Abor Assays, TX 77494, USA) following the manufacturer’s instructions. The data on morning corticosterone levels showed strong accordance between replicates and was therefore pooled for each mouse.
5.6 Gene Expression
5.6.1 Tissue Collection
Mice were sacrificed in the morning under isoflurane anesthesia by rapid decapitation. Organs were collected within 10 minutes after death and immediately stored in cooled RNAlater or for histology in 4% formaldehyde supplemented with 20% sucrose over night.
5.6.2 Generation of hiPSCs
Lines were derived from healthy patients genotyped for FKBP5 SNP rs1360780 and FKBP5 InDel rs9470080CNV. In both cases, the AT genotype corresponds to the ‘high induction’, or ‘risk’ allele, while the CG genotype corresponds to ‘low induction’ or ‘resilience’ allele. Lines were derived from peripheral blood mononuclear cells collected from 2 females and 2 males homozygote for the AT- or CG-allele, with even distribution of both genotypes. Reprogramming was performed with episomal plasmids [38]. Comparison of the genome wide CNV in the parental material and the emanated hiPSC showed no chromosomal aberrations. Pluripotency markers were detected immunocytochemically.
5.6.3 hiPSC differentiation
All hiPSC lines were cultured in mTSER1 (Stem Cell Technologies, Cat.No. 058509) on Matrigel Matrix High Concentration (Corning, Cat.No. 354263). Neural induction was performed based on a published protocol [39] with a few modifications. hiPS cells were maintained in Matrigel coated vessels, with mTeSR1 media and split by passing complete colonies using a non-enzymatic approach (EDTA, Versene Solution). hiPSCs were dissociated to single cells with Accutase (Stem Cell Technologies, Cat.No. 07920) and plated at 3 × 106 cells/well to allow the embryoid body (EB) formation in Neural Induction Media (NIM) + 10 μM Y-27632 (Stem Cell Technologies, Cat.No. 72308). They were allowed to attach for at least overnight, and then the medium was replaced by NIM (without Y-27632), consisting of a 1:1 mix of N2 supplement (Life Technologies, Cat.No. 17502048) in DMEM/F12 (Life Technologies, Cat.No. 31331028) and B27 supplement (Life Technologies Cat.No. 17504044) in Neurobasal (Life Technologies Cat.No. 21103049), supplemented with 10muM SB431542 (Millipore, 616461) and 1 μM Dorsomorphin (Tocris Bioscience, Cat.No. 3093). In days in vitro (DIV) 1-4, NIM was replaced twice a day. On day 4, the EB suspension were made and moved with a 5 ml serological pipette into a 6-well Clear Flat Bottom Ultra Low Attachment Multiple Well Plates (Corning, Cat.No. 3741), and cultured in NIM replaced daily for 10 days. On 10 DIV, the EBs were plated on tissue culture plates coated with Matrigel. On DIV 14-16, the neuroepithelial sheet was detached from the plate using STEMdiff Neural Rosette Selection Reagent (Stem Cell Technologies, Cat.No. 05832). From the following day until DIV 27 cultures were grown in Neuronal Maintenance Media (N2B27 supplemented with 20ng/μl hFGF) replaced daily or on alternate days. Between DIV 17 and 30, any non-neural differentiation present was removed by passaging with STEMdiff Neural Rosette Selection Reagent, and the neural cultures were then dissociated to single cells using Accutase. When cultures reached 80%–90% confluency, they were passaged again until a final passage between DIV 33-40, when they were plated for long-term culture, after which N2B27 medium was replaced every second day. At DIV 60 and 90, in each line, cells were detached with Accutase from several wells and filtered through 40 μm cell strainer (Corning, Cat.No. 352340) for FACS sorting evaluation.
5.6.4 FACS sorting
The media from 4 wells per line, containing astrocyte-neuron co-cultures were removed and the RLT buffer was added to wells for the RNA extraction. In parallel, at least 1 well per line (limited by the number of wells containing differentiated cells, variable between lines) was proceeded for separation of astrocytes and neurons from co-cultures using the positive selection approach with anti-CD44 (BD-bIoscience, Cat.No. 555478) antibodies [40]. Cells were gently detached from the well surface with Versene (3-5 minutes at 37°C), to avoid the epitope damage. Mechanical dissociation with p1000 pipette (5 times gentle up- and down strokes) was applied for obtaining single cells suspension. Cells were counted and up to 3×105 cells were incubated in flow cytometry (FC) wash buffer consisting of 1% FBS, 1× penicilin-streptomycin, nuclease free water and RNasin Plus RNase inhibitor 0.2 U/μl. Next, cells were incubated in the FC wash buffer (2h at 4 °C) containing FITC-coupled anti-CD44 antibody (1 to 80 BD Pharmingen Cat.No. 555478) or its isotype control (FITC Mouse IgG2b, k Isotype Control, BD Pharmingen Cat.No. 555478). After incubation, cells were washed in wash buffer (RNAase-free PBS, pH 7.4 + 0.2U/μl RNase inhibitor), spin down (300 g × 3’) and resuspended in 500μl of wash buffer. Separation was performed on FACSAria Machine (ZMBH, Heidelberg) at 4 °C using 100 μm nozzle (optimized for droplet stream). Based on side scatter pulse width and height (SSC-W and SSC-H), 200000-300000 events of singlets were sorted directly to Low binding tube coated with FBS overnight (Corning, CLS3207-250EA), spun down (400 g × 10’ at 4 °C) and the pellet was resuspended in 600μ of Qiazol Lysis Reagent (Qiagen, Cat.No. 79306) and kept at −80 °C before sequencing.
5.6.5 Next Generation Sequencing
RNA was isolated using RNeasy Plus kit (Qiagen Cat.No. 74192) following the manufacturer’s recommendations. RNA purity was checked spectrophotometrically using the NanoPhotometer (IMPLEN, CA, USA) and QIAxpert(Qiagen). Concentration was measured using Qubit RNA Assay Kit in Qubit 2.0 Flurometer (Life Technologies, CA, USA), while integrity was assessed using the standard sensitivity RNA kit (Cat.No. DNF-471, Advanced Analytical) on a Fragment Analyzer (Thermo Fisher Scientific, Langenselbold, Germany) and RNA Nano 6000 Assay Kit of the Bioanalyzer 2100 system (Agilent Technologies, CA, USA). High quality RNA samples with RIN >7.5 were eligible for further processing.
A total amount of 1 μg RNA per sample was used as input material for the RNA sample preparations. Enrichment of mRNA from eukaryotic organisms was performed using oligo(dT) beads from NEBNext Poly(A) mRNA Magnetic Isolation Module (Cat.No. E7490L, NEB, USA). Subsequently, sequencing libraries were generated using NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (Cat.No. E7770L, NEB, USA) following manufacturer’s recommendations. Briefly, fragmentation was carried out using divalent cations under elevated temperature in NEBNext First Strand Synthesis Reaction Buffer (5×). First strand cDNA was synthesized using random hexamer primer and M-MuLV Reverse Transcriptase (RNaseH-). Second strand cDNA synthesis was subsequently performed using DNA Polymerase I and RNase H. In the reaction buffer, dNTPs with dTTP were replaced by dUTP. Remaining overhangs were converted into blunt ends via exonuclease/polymerase activities. After adenylation of 3’ ends of DNA fragments, NEBNext Adaptor with hairpin loop structure were ligated to prepare for hybridization.
In order to select cDNA fragments of preferentially 250-300 bp in length, the library fragments were purified with AMPure XP beads (Cat.No. A63987 Beckman Coulter, Beverly, USA). Then 3μl USER Enzyme (NEB, USA) was used with size-selected, adaptor-ligated cDNA at 37°C for 15 min followed by 5 min at 95°C before PCR. Then PCR was performed with Phusion High-Fidelity DNA polymerase, Universal PCR primers and Index (X) Primer. At last, products were purified (AMPure XP beads) and library quality was assessed using the Agilent High Sensitivity DNA Kit (Cat.No. 5067-4626) on the Agilent Bioanalyzer 2100 system (Agilent Technologies, CA, USA).
The clustering of the index-coded samples was performed on a cBot Cluster Generation System (Cat.No. SY401-2015, Illumina) using TruSeq PE Cluster Kit v3-cBot-HS (Cat.No. PE-401-3001, Illumia) according to the manufacturer’s instructions. After cluster generation, the libraries were sequenced on a NovaSeq 6000 Illumina platform using NovaSeq 6000 S2 Reagent Kit v1.5 cat. 20028314 -(300 cycles) and 150 bp paired-end reads were generated (minimum 12 Gb and 40 M).
RNA-Seq reads were aligned to the mouse genome using Hisat2 software, version 2.1.0 with the corresponding Ensembl GRCm38.p6 reference genome (http://www.ensembl.org). Confirmation of genotyping was done by aligning the NGS reads to the human reference genome GRCh38.p13. Sequenced read quality and duplications were checked with FastQC software, version 0.11.9 and alignment quality metrics were calculated using Samtools flagstat software, version 1.10. Gene and transcripts expression profiles were quantified using Cuffquant and Cuffnorm, version 2.2.1 and GTF file from the Ensembl (v. 100) database to obtain Fragments Per Kilobase Million mapped reads (FPKM). Library preparation, sequencing and initial data processing was carried out at Intelliseq (Poland).
5.7 Statistical Analyses
Data processing and analysis was carried out using R (version 4.0.2). For analysis of read outs with repeated measurements (open field, dark-light and social chamber test, HPA-axis performance), nested models using early life condition group, mouse strain and sex as between factors and compartment or time point as within subject component were defined and their quality was inspected visually and using one-point cross validation (R packages nlme and afex). Data without a temporal or spatial component were modeled linearly. Confidence intervals for the coefficient estimates were obtained using the non-centrality parameter method and the Greenhouse-Geiser method for approximation of the degrees of freedom was applied for nested models [41]. Analysis of variance and effect size estimation were performed using partial sum of squares type II (R packages car and effectsize). Besides the generalized eta squared (gη2), the partial epsilon squared (pϵ2) effect size were reported to reduce potential bias by small sample size [42, 43, 44]. In addition, the relative explanation of variance was assessed (R packages MuMIn and r2glmm). If significant model terms were suggestive, pairwise two-sided post hoc tests with Tukey contrasts were performed (package emmeans). Effect sizes in the descriptive analyses between subgroups were computed as Cohen’s d estimates using pooled variance.
Regarding data obtained from next generation sequencing, the following additional analysis steps were performed: The obtained data was filtered for tissue-wise median and mean expression to be above 1 FPKM. In addition, selection criteria for fold changes and signal-to-noise ratios bigger than ± 30% and ± 1.5 were applied, respectively. Comparisons were made between control and ELA-exposed females independent of strain as well as separate for the subgroups of AT- and CG-allele carriers. Furthermore, putative differences between AT- and CG-allele carriers independent of early life experiences, and within the control and maternally separated subset were investigated. To all transcripts where the row-wise t-test was significant, a false-discovery rate filter of 10% was applied and only genes of which the related transcripts indicated fold changes in the same direction were considered as differentially expressed genes (DEGs) in subsequent analyses. Based on the association of the CG-allele with resiliency and the AT-allele with risk to develop disorders, DEGs which were unique to CG-allele carriers when comparing effects of early life conditions (coping) were labeled as potentially resiliency-associated genes, while an overlap of DEGs from the early life comparison with DEGs from the SNP-comparison in the control subgroup were labeled as potential vulnerability-related genes. In addition, transcripts were the 2-way ANOVA suggested an interaction of early life condition × Fkbp5-genotype at an α level of 5% were included in gene set enrichment analyses.
Using Ingenuity (IPA, Qiagen), the FPKM values of the listed DEGs were subjected to ‘core analysis’ of ELA vs. control for the AT- and CG-allele carrying subgroup applying a threshold of absolute fold changes bigger than 1.5. The results of these ‘core analyses’ were entered into a ‘comparison analysis’ to investigate the deregulation of genes due to ELA between strains. The ‘comparison analysis’ was limited to the term ‘Diseases and Biological Functions’, and further limited to sub-categories of neurological relevance. The z-score p-value was set to <0.0001. The final comparison list was then filtered to the top 10 results, ordered by z-score.
Pearson correlations were computed tissue-wise for the normalized expression levels of Fkbp5 and the DEGs with HPA-axis and behavior-related read outs. The cutoff for meaningful correlations was a priori set to > |0.6|. Among those correlations, the FDR was fixed to 5%.
Generally applicable gene-set enrichment analyses (package gage) for metabolic pathways listed in the Kyoto Encyclopedia of Genes and Genomes (KEGG) were performed and visualized in case of significant over representation (package pathview). For comparison to the hiPSC-derived astrocytes and neurons, the genes listed in the circadian entrainment pathway (hsa004713) were extracted, filtered based on the above-mentioned effect size and significance criteria and visualized without prior checks on pathway enrichment due to the limitations in sample size.
6 Accompanying Statements
6.2 Author Contributions
Allers, KA: conceptualization, maternal separation, data interpretation, manuscript revision
Blasius, A: T maze
Del Prete, D: cultivation, FACS and NGS of hiPSCs
Harris, I: support RNA isolation
Hengerer, B: conceptualization, revision of the manuscript
Kolassa, IT: data interpretation, revision of the manuscript
Koros, E: support with maternal separation
Nold, V: conceptualization and execution, sample and data analysis, manuscript
Peleh, T: support social chamber test
Portenhauser, M: support RNA isolation, corticosterone assay, social chamber test
Slezak, M: conceptualization hiPSC experiments, manuscript revision
6.3 Competing Interests Statement
Isabella Harris and Iris-Tatiana Kolassa declare no conflict of interest. Kelly Ann Allers, Andrea Blasius, Bastian Hengerer, Eliza Koros, Verena Nold, Tatiana Peleh and Michelle Portenhauser are employees at Boehringer Ingelheim Pharma GmbH & Co KG. Michal Slezak and Dolores Del Prete were employees at BioMedX during preparation of data used in this publication.
The funding for this study was provided by Boehringer Ingelheim Pharma GmbH & Co KG to provide a doctorate thesis project to Verena Nold. The company had no further influence on this work.
6.4 Ethics Statement
Experiments were performed under the allowance of the regional council for animal welfare (Regierungspräsidium Tübingen, Baden-Württemberg, Germany) and adhere to ARRIVE guidelines.
6.5 Data Availability Statement
All raw data files and processed summary data frames will be made available by the corresponding author upon request. The accession code for the murine NGS data set on NCBI’s Sequence Read Archive is PRJNA743189. The hiPSC data sets will be made available upon request.
6.6 Code Availability Statement
All R code written to process, analyze and visualize the herein contained data will be made available by the corresponding author upon request.
7 Supplementary Material
6.1 Acknowledgements
The authors wish to thank their colleagues at Boehringer Ingelheim Pharma GmbH & Co KG Margot Weiland, Sonja Diehl, Nadine Richter, Marion Trautmann, Werner Rust, Birgit Stierstorfer, Tanja Schönberger for their excellent support in processing of the obtained samples as well as Silke Laack-Reinhardt, Yvonne Schneider, Sonja Hofbauer, Ralf Weber, Britta Gerth and Lukas Schmidt for their help with breeding of the animals. In addition, the hiPSC work would not have been possible without the exceptional support of Susanne Zach (Boehringer Ingelheim Pharma GmbH & Co KG), Shringarika Singh and Santiago Tena (BioMedX). We furthermore thank Michal Korostynski, Slawomir Golda, Dzesika Hoinkis and Marcin Piechota at In-telliseq for carrying out the next generation sequencing and discussing the thereof obtained data. Special thanks go to Michael Schuler for his support during the realization of the novel mouse lines together with Susie Mikkelsen at Taconic Biosciences. Lastly, the authors are greateful for the support of Elisabeth Binder at the Max Planck Institute for Psychiatry in Munich for her guidance in the conceptualization of the humanized mice and the provision with hiPSCs from patients and healthy controls.